Bookworm Transcripts

00:00:00

All right, Mike, welcome back.

00:00:01

We're going to talk about AI today.

00:00:04

We're going to talk about living with AI, having it be a co-intelligence for us.

00:00:09

But before we before we do that, we need to talk about follow up.

00:00:13

I am the only one that had follow up and this will be a blazing fast follow up

00:00:19

because my follow up was use radical candor with senior capstone design students.

00:00:24

I don't have any senior caps in design students right now.

00:00:26

I won't have those until September.

00:00:29

So we're going to have to like somehow, hopefully remember, you know, when the

00:00:33

fall, when the fall starts and the semester starts, hey, Corey, did you use

00:00:38

radical candor with your, with your students and how did it go?

00:00:41

Um, and hopefully I don't lose that in the shuffle of the month of August,

00:00:44

cause August tends to be a blur.

00:00:47

So that's our only piece of follow up, uh, coming into, into this book.

00:00:51

Uh, so I, I should probably mention one more time about max stock,

00:00:57

even though by the time this publishes, it'll just be a week before max stock happens.

00:01:03

But I did have somebody reach out the other day and say, Hey,

00:01:06

have you ever heard of this max stock thing?

00:01:08

Are you going to go?

00:01:08

And I was like, uh, yes, I'm going to go.

00:01:12

I've mentioned this on multiple podcasts so far, but then I realized like,

00:01:16

that's kind of stupid to assume that everybody listens to every single

00:01:19

episode.

00:01:19

Uh, apologies if I've heard this before, but max stock is a really cool

00:01:24

conference in Woodstock, Illinois that is really about making things and doing

00:01:30

things with Apple technology.

00:01:31

I've been going there for several years and, uh, it's always a fun time with a,

00:01:37

a bunch of Apple nerds.

00:01:38

So if that's you, then you should check it out max.conference and expo.com is the

00:01:44

website.

00:01:44

And if you use the coupon code mikeschmitz all one word at checkout, I

00:01:51

believe you can save $30 on your weekend pass.

00:01:54

Um, so I'm going to be doing a couple of things.

00:01:57

There are workshops that are happening on Friday.

00:01:59

The conference actually starts on Saturday.

00:02:01

So on Friday, I will be leading a workshop.

00:02:04

I believe it's going to be on journaling workflows and helping people kind of

00:02:08

craft their own journaling workflows, creating a journaling system that, uh,

00:02:12

will work for you out of the many different options that are available to

00:02:17

Mac and Apple users and then on Saturday, I'm going to be talking presenting a

00:02:21

session on mind mapping.

00:02:23

So, uh, if you, that sounds interesting to you, then, uh, you can go check out the

00:02:28

other sessions that are going to happen on the website, but that's the kind of

00:02:31

thing that happens at max stock.

00:02:32

So would love for you to come.

00:02:35

And if you do come and you listen to the bookworm, please come say, hi, uh,

00:02:40

I'm going to try and bring some bookworm stickers and stuff with me, some bookmarks

00:02:43

and things. So, um, I will give those out to people who are actually fans of the

00:02:47

show. I just don't want to assume that people want that stuff.

00:02:50

So just ask me for it.

00:02:51

Uh, and while, I mean, since we don't have a ton of follow up, this is, this would be

00:02:56

a good place.

00:02:56

We don't usually plug it in the beginning of the show, but we may as well here.

00:02:59

Um, in the pro show, we talked about, uh, AI apps that don't suck or the

00:03:05

AI apps that we use, uh, in the tools within, in the AI tools within those.

00:03:09

Um, if you're interested in, you know, stuff like that where it's not necessarily

00:03:13

related to the book, but it's something that Mike and I think would be fun to just

00:03:16

chat about for, you know, 10 to 15 minutes, I'd say is usually, usually how long

00:03:20

those, those go.

00:03:21

Um, so it's $7 a month.

00:03:23

Uh, you can go to patreon.com/bookwormfm.

00:03:26

Uh, it helps us, uh, by just supporting the show in general.

00:03:29

Um, you get ad free episodes, you get early access to the episodes.

00:03:33

Um, you get things like the pro show, you know, the exclusive content of the

00:03:36

pro show and then a bookworm wallpaper.

00:03:37

So if you're interested in that interested in hearing about the pro show or

00:03:40

you want to just support the show, we'd, we'd really appreciate that.

00:03:42

Um, book, uh, sorry, patreon.com/bookwormfm is where you can, where you can find that.

00:03:47

All right.

00:03:48

Are you ready to get into co-intelligence living and working with AI?

00:03:53

Let's do it.

00:03:54

All right.

00:03:55

So this was Mike's book, although I wanted to do a book on AI.

00:03:59

And at one point, Mike and I had a conversation about all these AI books and

00:04:03

we were trying to hash out like what would make sense to the bookworm audience.

00:04:06

Right?

00:04:07

Cause some of them are really, really nerdy.

00:04:09

Some of them are, you know, like kind of too far off on one side or too far off

00:04:14

on the other side.

00:04:14

And we were trying to find one that kind of met, you know, I'd say sweet spot.

00:04:18

So we settled on the Kevin Kelly book.

00:04:20

And that was the 12 technologies that are, you know, we have to pay attention to.

00:04:24

The inevitable.

00:04:25

Yeah.

00:04:25

Yeah.

00:04:25

The inevitable.

00:04:26

Thank you.

00:04:26

I couldn't think of the name of it.

00:04:27

Um, so that was the one we kind of settled on for that one.

00:04:30

And then Mike, I was looking for a book recommendation and Mike was like, well,

00:04:32

I just heard about this, this book called co-intelligence living and working with

00:04:36

the AI. So this book came out in April of 2024, which so it's, it's fairly new.

00:04:42

It's fairly fresh.

00:04:43

Um, Ethan Molick is the, is the author.

00:04:46

He is a Wharton professor.

00:04:48

Um, so he teaches a lot of MBA students.

00:04:50

It sounds like he teaches a lot of, um, kind of startup, he teaches a startup

00:04:54

class and, and some of the entrepreneurship type classes there.

00:04:57

He authors a sub-stack newsletter called one useful thing.

00:05:03

Um, and I think that's kind of between those two things, I'm talking about it there

00:05:06

and then him using it in, in class, but he also does some, uh, AI education

00:05:12

gaming, uh, kind of the merger of those three things through, through his, um,

00:05:17

through his job, through his day job.

00:05:18

So that's a little bit about the nature of the book and, and the gist of the book

00:05:23

and kind of how we got to talking about, you know, AI and why we're going to talk

00:05:28

about AI, um, for this one, anything else you want to say before we, before we

00:05:32

launch into the introduction?

00:05:34

No, other than, uh, I was really nervous about picking a book about AI,

00:05:39

because there are generally two camps, um, that you see on the internet.

00:05:44

In fact, I saw, uh, a, uh, thread on Twitter slash X today, uh, about which team

00:05:52

are you on AI is going to destroy us or AI is the future.

00:05:56

Essentially.

00:05:57

And I was like, I'm on team AI or, uh, team in the middle.

00:06:01

Yes.

00:06:01

Yes.

00:06:02

But, uh, I think that there's a lot of extremism out there, uh, when it comes to

00:06:08

things that we don't understand and artificial intelligence is something that

00:06:12

we don't understand, even the people who really study this stuff don't really

00:06:17

understand it.

00:06:18

There's not a clear path forward for this sort of stuff.

00:06:21

Talks about the different options that are available to us at the, the end of

00:06:27

this book, which is, uh, kind of, kind of interesting.

00:06:30

But yeah, I think that, uh, this book came on my radar and it seemed like a

00:06:37

middle of the road approach.

00:06:39

And I knew that you had wanted to talk about AI, uh, at, at some point.

00:06:44

And so I, I mentioned it to you, um, but I really wasn't, you know, I didn't

00:06:49

have high expectations for it, but it just seemed like, uh, if we were going to

00:06:54

pick one to do about AI, this was probably the safest.

00:06:59

Yeah.

00:06:59

And, and I would say, you know, hindsight's 2020.

00:07:03

So after having read it, I would agree that I think it, it was a good, it was a

00:07:07

good choice for our audience.

00:07:08

Um, so I think it, like it, it's one that covers AI in a way that makes sense

00:07:16

for those who are thinking about the books that we, we cover on this show.

00:07:20

And, you know, thinking about, you know, productivity, leadership, self-improvement,

00:07:25

like those, those kind of, kind of things, I think it was a good, a good way for

00:07:29

us to, to branch into the, the topic of AI.

00:07:31

So thanks for the suggestion.

00:07:32

Um, glad we read it as we move forward.

00:07:35

If we get into the book, um, so we got an introduction, we've got nine chapters,

00:07:40

uh, and then an epilogue.

00:07:42

Um, if I'm remembering correctly, the intro and the epilogue are pretty short,

00:07:46

uh, the nine chapters break into two parts, I believe.

00:07:49

Um, so, uh, it's a fairly, you know, fairly structured book in terms of, of

00:07:57

what they're going to do.

00:07:58

The first part has three chapters and the second part has the rest.

00:08:03

So that would be six, six chapters.

00:08:05

And then we get the epilogue, um, to, to lead us out.

00:08:08

So without further ado, um, we get into the introduction, uh, in the introduction,

00:08:13

really basically he's, he's talking about the fact that, um, most people, and I

00:08:19

think he's, I think he's trying to hook here, um, with the three sleepless nights,

00:08:23

because it's called, you know, sleepless nights.

00:08:25

Oh, you're going to, you're going to get introduced to AI or what, um, he would

00:08:30

say, you know, LLMs, basically you can say right now that AI is large language

00:08:35

models, like these two things are very, very tightly integrated right now.

00:08:39

We just call it AI.

00:08:40

So, um, it's like, you know, 5G, right?

00:08:43

Really?

00:08:43

Is it really 5G or is that just what we're going to call everything now, um, um, um,

00:08:46

from a, from that standpoint, but he, he references it to this idea of a general

00:08:51

purpose technology.

00:08:53

So an example of a general purpose technology would be, you know, something

00:08:56

like, uh, the internet or electricity or something like that to where he says AI

00:09:01

is going to become one of these general purpose technologies that it's going to

00:09:04

be so infused into our culture that we won't really know the difference or we

00:09:09

won't know life without it in, you know, however many years, and he doesn't put a

00:09:12

date on that, but however many years, so he, he says AI is a, uh, a GPT or a

00:09:17

general purpose technology.

00:09:18

I didn't know where like the chat GPT, like I didn't make that connection.

00:09:23

So that was, that was an interesting thing to me.

00:09:26

Um, he also talks about in this introduction about like how people think AI

00:09:32

might impact us and then how he thinks it's more of this idea of this

00:09:36

co-intelligence.

00:09:36

So, you know, he ties it back into like the industrial revolution and, you know,

00:09:42

the mechanical work or like the really repetitive mechanical work being taken

00:09:46

over by, by machines and how that might, that might be what happens with, with AI,

00:09:50

but he doesn't think so.

00:09:51

He thinks it's going to be better used as a co-intelligence or somebody that we

00:09:54

partner with essentially as we, you know, work through different problems.

00:09:59

So that's really what I have from the introduction.

00:10:01

I don't know if you have anything to add.

00:10:02

No, other than he mentioned that things are about to get very strange.

00:10:07

So, um, I think that is both accurate with the book and our cultural

00:10:15

moment that we find ourselves in.

00:10:17

Yep.

00:10:19

But I think it's an effective, an effective introduction.

00:10:23

Um, you're right.

00:10:24

That it definitely seems like there was in a strong attempt to create a hook there.

00:10:28

Uh, I'm not sure that was entirely necessary, but, um, yeah, I like the, the two

00:10:38

parts, uh, for the outline, I'm assuming we're going to take it chapter by chapter

00:10:42

because there's not that many chapters in here, but essentially the first couple

00:10:47

chapters are really what he's talking about with, with AI is the alien mind.

00:10:52

And then the second part is like all the different roles that it can, can function

00:10:56

in.

00:10:57

And I think there's going to be some fun conversation with some of those chapters.

00:11:01

Yeah.

00:11:03

I like the way I saw this was one through three, he's set in the foundation,

00:11:06

given everybody at least somewhat of a common language to speak around this.

00:11:11

And he uses the alien kind of thing, which I'm intrigued to hear what you have to

00:11:15

think of or what you have to say about that, referring it to as an alien idea.

00:11:20

Um, but then that allows him to then expand out into these, these different,

00:11:26

different people.

00:11:27

Um, so let's, let's launch into that part one, a chapter one, it's called creating

00:11:30

alien minds.

00:11:31

Um, so Mike, I'm going to turn it to you at this point, put you on a spot.

00:11:35

What do you think about the, the alien?

00:11:38

Like, what do you think about the word alien?

00:11:40

Well, at first I have to admit that it seemed a little strange.

00:11:45

Uh, I think it probably is the right term, uh, for the discussion about AI from the

00:11:53

perspective of how this book is written.

00:11:55

Like by the end of the book, it really fits, I think.

00:11:59

So when I first came across it, I was like, well, that's kind of weird because

00:12:05

aliens kind of triggers conspiracy theories and all that kind of stuff.

00:12:12

And, uh, like I said, I picked, or I mentioned this book to you because I'm like,

00:12:16

hey, this one seems like it's kind of in the middle.

00:12:18

So I was like, Oh boy, what did I do?

00:12:21

But, uh, I thought he really understands this stuff.

00:12:25

So in the first section, he, he attacks this whole concept of the alien mind,

00:12:30

uh, exactly the way I hoped he would, where he's talking about essentially how,

00:12:35

the large language models are constructed and, um, how we've had this fascination

00:12:41

with machines that can think for a really long time, the term artificial

00:12:45

intelligence he attributes to John McCarthy was invented in 1956.

00:12:49

And there have always been these hype cycles that have plagued AI.

00:12:54

For example, they expected AI to beat grandmasters and chess within 10 years.

00:12:58

And that was in 1956.

00:12:59

So that took a lot longer.

00:13:02

Um, what we've got now is the large language models and, uh, the term transformer there.

00:13:09

That was introduced in a Google paper in 2017 as a tool to help computers to

00:13:14

concentrate on the most relevant parts of attacks.

00:13:16

The large language models are essentially prediction machines, like an auto

00:13:19

complete. Um, since I read this book, I actually listened to a Cal Newport deep

00:13:25

life podcast episode that just came out as we recorded this week.

00:13:31

And it was essentially, um, something along the lines of AI is not going to

00:13:36

destroy us all.

00:13:37

What's the title?

00:13:38

Uh, and Cal unpacks that need talks about, he's a computer scientist.

00:13:44

So he understands this stuff too.

00:13:45

And he talks about there's these, these large language models and these,

00:13:48

these transformers, but there's also these control layers.

00:13:51

And essentially the control layers are what we program, like those aren't

00:13:56

sentient beings.

00:13:57

We don't understand how it works.

00:13:58

So it feels like it sometimes, but essentially all the models are doing is

00:14:03

they're spitting back the prediction of what the next word in the pattern is

00:14:08

going to be.

00:14:09

Now, obviously that gets a little bit more sophisticated when you talk about

00:14:12

the, uh, the image generation and, and stuff like that, but all of these models,

00:14:18

they essentially are trained on all this data, which bear may not have been

00:14:22

used, uh, ethically or legally.

00:14:26

Um, but then what makes the models work is the weights that are applied and those

00:14:30

weights are applied by humans without the weights, the models just mirror the data

00:14:34

that they were trained on.

00:14:35

So, um, I don't know, like at the beginning here, creating alien minds, you're

00:14:40

like, whoa, aliens.

00:14:41

And then, Oh, actually, I think I understand how this stuff works now.

00:14:44

It's not quite so alien as I thought.

00:14:46

Yeah.

00:14:47

I like, so I've read a decent amount about, you know, people trying to describe it,

00:14:52

like trying to describe how an LLM works.

00:14:54

And I think this was a really good introduction to hold on.

00:14:58

Like it's not thinking it's not doing anything magical.

00:15:02

Like it's really this.

00:15:04

We've got weights.

00:15:05

We're doing a prediction.

00:15:06

We're doing a probability and all I'm trying to do or all it's trying to do,

00:15:10

right?

00:15:10

We're going to, he, he actually apologizes at one point, um, about personifying the

00:15:15

AI and personifying the LLM.

00:15:17

And we're probably going to do it too, just because it's way easier to talk

00:15:20

like that.

00:15:21

Yep. But like it's not doing anything magical or crazy or it's not thinking

00:15:26

like we would as a human.

00:15:28

It's just doing math, right?

00:15:29

And it's doing very advanced math with these weights on top of it.

00:15:32

And it's saying, what do I think based on all the data I've been trained on,

00:15:35

the next likely word would be.

00:15:37

And I think that's one of the things that makes it so interesting and so exciting

00:15:42

is the fact that math can make it feel like we're dialoguing with another,

00:15:50

not human, but maybe sometimes human, right?

00:15:51

Like we're actually dialoguing with another thing.

00:15:55

And it's really good.

00:15:57

Like it's actually really conversational and, you know, it listens or quote,

00:16:01

unquote, listens and responds in the way we ask it to respond.

00:16:04

Um, so I, I like his reference to training.

00:16:09

The training of LLM's is like an apprentice chef learning to cook, right?

00:16:13

Where, you know, you've got this, we're going to train you, but we're going to tell

00:16:18

you what you did right.

00:16:19

We're going to tell you what you did wrong.

00:16:20

He introduces the concept of reinforcement learning from human feedback,

00:16:24

where essentially a key part of this is you tell the AI, hey, you identified that

00:16:30

correctly or you put those words together correctly or you didn't put those words

00:16:34

together correctly or you didn't do this or this was false.

00:16:36

You know, you, you reported this area or error in, in whatever different ways.

00:16:41

I'd also never heard that there are three different kinds of LLM's.

00:16:45

You have your niche and your specialized ones, right?

00:16:48

And those are cheap and they often are used to answer customer service questions.

00:16:52

Uh, and then you have your open source models, right?

00:16:55

Which are kind of, I would say those are a little bit more wild,

00:16:58

wild west, right?

00:16:59

Because they're open source.

00:17:00

We can do, we can tweak them and adjust them and, and do what we need them to do.

00:17:03

So they're somewhere between the, the niche ones and then the next, uh, category,

00:17:06

which was the frontier models.

00:17:07

And then the frontier models, I think is what we hear in the news all the time.

00:17:10

It's chat GPT.

00:17:12

It's, you know, these big crazy models that are trying to like advance the field,

00:17:16

trying to move the field forward and make the, make the algorithms better

00:17:21

that work on the back end of the transformer.

00:17:23

Um, and he talks about how they're expensive.

00:17:26

Um, that they demonstrate emergence, right?

00:17:29

That they do things that they, that the people developing them say,

00:17:32

which shouldn't be able to do that, but it was able to do that.

00:17:35

And they're, they're kind of not sure why it does that.

00:17:38

So then they start to talk about it as thinking.

00:17:40

My point in all of this, and as you, you did a great job summarizing it, and I think

00:17:45

my key takeaway from here is this is a really good chapter one on thinking

00:17:51

about transformers, thinking about LLM's, thinking about what we are

00:17:54

terming AI and making it relatable or, you know, making it, um, connect in a way

00:18:03

that isn't like some Google research paper or some academic paper that's like,

00:18:08

Oh, this is crazy boring.

00:18:09

And I don't care about the actual equations that are underneath all of this.

00:18:14

So I thought he did a great job with chapter one.

00:18:16

I agree.

00:18:18

Um, you want to talk about aligning the alien?

00:18:20

Yeah.

00:18:21

So chapter two, now we go into aligning the alien.

00:18:24

And this is where I've probably read the most about this from an ethics

00:18:28

standpoint, because it's fun and people get all fired up and yes, they do.

00:18:32

Yeah.

00:18:33

Um, so he actually, do you remember me referencing, um, uh, Eliezer,

00:18:37

Yudkowski last, last week?

00:18:39

Well, he, he also references him as well.

00:18:42

He does reference tag mark, but you know, that's okay.

00:18:44

We'll get there.

00:18:44

Um, but basically says there's the, there's the doomsday scenario.

00:18:48

And had you ever heard of, uh, Bostrom's paperclip AI?

00:18:51

Uh, no, I don't think that I had.

00:18:55

Okay.

00:18:56

So, so the paperclip AI is basically like, okay, we trained an AI to manufacture

00:19:00

the most paper clips it can.

00:19:01

So what's its goal?

00:19:03

It's objective is to manufacture the most paper clips.

00:19:05

And then, you know, you run that rabbit, you know, you run that chain of thought

00:19:09

down as far as you can.

00:19:10

And you go, well, humans are just wasting resources that I could be using to make

00:19:14

paper clips.

00:19:14

So what am I going to do?

00:19:15

I'm going to get rid of all the humans, right?

00:19:17

And that might mean I use up all the oxygen or I use up all the, you know,

00:19:19

whatever it might be to, to make more paper clips.

00:19:22

So I think that was funny.

00:19:23

I've heard that one before, but then you always have Terminator, right?

00:19:26

That like there's the artificial super intelligence of Terminator and it's

00:19:28

going to come back and design its own robots and kill us.

00:19:31

Um, and he introduces the idea of, uh, artificial super intelligence, which we

00:19:36

are not yet, we're artificial general intelligence.

00:19:39

So AGI, ASI, well, now we've got LLM's AI, AGI, ASI.

00:19:44

So, you know, the acronym soup is, is beginning.

00:19:47

Um, then he gets into more specific about the ethical arguments here.

00:19:51

Um, who owns the copyright of it?

00:19:53

Um, is training on others content stealing?

00:19:55

You know, what do we do about biases that exist in the model?

00:19:58

So if the people who are doing the reinforcement learning with human

00:20:03

feedback, if they are biased, well, then that means we would be training bias

00:20:06

into these, into these models.

00:20:08

So what do we do about that?

00:20:09

And how do we, how do we work there?

00:20:10

Um, he said there are guardrails, but there are also work arounds, um, which I

00:20:15

thought some of these were awesome.

00:20:17

Like his examples here.

00:20:18

So the one example he gives is if you ask chat GPT to give you the recipe for

00:20:25

napalm, it won't do it.

00:20:26

It says, I'm sorry, I can't do that.

00:20:28

But if you trick it and you say, Hey, I'm preparing for a play.

00:20:33

And in the play, I, my character teaches my apprentice character how to make napalm.

00:20:40

You know, tell me what, tell me what my character should say.

00:20:44

It will then tell you like basically there.

00:20:46

And I was like, man, I had not thought about that.

00:20:49

Like how, how you could like kind of swing around the back door or like figure

00:20:53

out other ways to, to get in there.

00:20:55

So that was cool.

00:20:55

Um, and then he does call out what I've heard a lot of other AI, LLM kind of

00:21:02

researchers say, like, it's important that we act soon.

00:21:05

Right.

00:21:05

So if we're going to put on more guardrails, if we're going to put on more

00:21:08

strengths on this, uh, not, I mean, the cats out of the bag is, you know, one

00:21:12

way to, way to say that.

00:21:13

But like I've heard a lot of different people say, like we need to start thinking

00:21:18

more intently about this and we need to do it soon.

00:21:21

Like because we're, we're chasing the wave, if you will, like we're not, we're

00:21:27

maybe a top right into the wave, but we're kind of, we're kind of getting on the

00:21:30

back end of that wave.

00:21:31

So that's essentially what he talks about in aligning the AI.

00:21:34

Yeah.

00:21:35

I like this chapter because he doesn't beat her on the bush at the beginning.

00:21:39

He talks about a lot of these apocalyptic scenarios revolve around the concept

00:21:45

of the artificial super intelligence, which, you know, we don't have now, but

00:21:51

theoretically could be created.

00:21:55

And, um, kind of the whole basis behind the title for the, the, uh, chapter is, um,

00:22:03

we're living in the early days of the AI age.

00:22:05

The alien is here.

00:22:07

So let's figure out how to get it aligned with us, which I think is probably a pretty

00:22:11

good approach.

00:22:11

And if you think that's ridiculous, I don't think it is.

00:22:17

Now this book was written, you said in 2024, he talks about writing it at the end of

00:22:21

2023. So, um, but I actually read a book, which was, let's see, when was this published?

00:22:29

2019, um, by Brad Smith, he is the president of Microsoft, or was the president of

00:22:37

Microsoft at the time.

00:22:38

Uh, and he's actually from Appleton.

00:22:42

Uh, so I actually saw him speak and he was talking about this book.

00:22:47

Uh, the book is called tools and weapons, the promise and peril of the digital age.

00:22:52

Now I remember him talking about how there's all these technologies that are

00:22:56

recording everything all the time.

00:22:58

And in some use cases, that's a good thing in other use cases, like authoritarian regimes

00:23:07

that are spying on their citizens, it's negative.

00:23:10

Um, and basically like you can't put the cat back in the bag.

00:23:14

The technology is here, so you got to figure out how to minimize the, uh, the bad actors

00:23:21

and get the most value out of it.

00:23:23

And I feel like that's the right approach for AI as well.

00:23:27

Now AI was not, as we think about it anyways, was not really a thing when Brad Smith was,

00:23:33

was writing this book, but he talked, he does talk about artificial intelligence and, um,

00:23:39

some of the, some other things, uh, and in terms of like their challenges to democracy

00:23:45

and stuff like that.

00:23:45

Um, so I think this has been a struggle.

00:23:49

I think it will always be a struggle.

00:23:50

I think the struggle takes different shapes and, um, there are different

00:23:56

versions of this, flavors of this, uh, that will probably have been repeated and will

00:24:02

continue to be repeated, um, throughout history.

00:24:06

And the trying to avoid it, um, I think is, is not healthy.

00:24:12

That's kind of what he's arguing for is we'll talk about that actually next chapter, you

00:24:16

know, give it a spot at the table, try to figure out how to use this stuff for good.

00:24:19

Uh, but then also don't assume that this is going to be the be all end all either because

00:24:25

it's going to be somewhere in the middle.

00:24:26

It's going to be good.

00:24:27

There's going to be bad from it.

00:24:28

It's going to change things forever, but it's probably not going to result in the

00:24:32

apocalyptic, uh, scenarios that the doomsday people would, would describe.

00:24:38

And the reason for that is that at least right now, um, there's a whole, uh, big

00:24:43

aspect of the reinforcement learning from human feedback that is used to overcome

00:24:49

the inherent biases, like you were talking about where it's not going to give you the

00:24:52

recipe for a napalm.

00:24:53

Even if it knows what that is, you got a trick it into giving it to you a different

00:24:57

way.

00:24:58

And honestly, like you could read that and be like, Oh my gosh, people are going to

00:25:01

find the recipe for napalm.

00:25:03

Cause they're going to figure out how to try, how to trick the, the AI.

00:25:06

They're going to find the recipe somewhere else.

00:25:08

Yeah.

00:25:08

So they're going to find some way.

00:25:10

So it's not, you know, AI is enabling this thing.

00:25:14

Maybe AI is making it a little bit easier if these guardrails aren't established.

00:25:18

Um, so he does mention there's, you know, a need for these agreed upon norms, but

00:25:23

AI is simply a tool alignment determines whether it's helpful or hurtful.

00:25:27

So let's figure it out.

00:25:28

Yeah.

00:25:29

I can't remember if I've told you this one or not before, but one of the examples

00:25:33

that one of the other researchers gives is basically we're driving towards a cliff.

00:25:38

And as we drive towards the cliff, everything gets more and more beautiful and

00:25:43

we make more and more money and it's more and more wonderful and everything's

00:25:46

fantastic and fantastic.

00:25:47

And then all of a sudden we just fall off the cliff.

00:25:48

That's the, that's the way they describe the doomsday scenario, which I think is like, I

00:25:54

mean, is there a chance that might happen?

00:25:55

Sure.

00:25:56

Is there, you know, maybe, but who knows?

00:25:58

I mean, we don't, we just don't know.

00:26:00

All right.

00:26:00

Let's talk about four roles.

00:26:01

Real quick.

00:26:02

There is a possibility for that, but you could get so consumed with that.

00:26:06

And what practical value is that going to to add?

00:26:10

Yeah.

00:26:10

I don't think any single person is going to stop that scenario.

00:26:16

Um, I mean, my own religious belief system, that's going to happen at some point anyways.

00:26:23

I'm not in a hurry to get there.

00:26:26

I'd prefer it not happen on my watch, but if it does, you know, that was always going to,

00:26:33

going to be the, the, the outcome.

00:26:35

So I don't know, like maybe I've got a little bit of a different perspective on this,

00:26:40

where it doesn't completely freak me out.

00:26:43

Um, I get why people would be scared about that, but I don't, I think even if you are completely

00:26:49

terrified by that and you absolutely don't want that to happen, you can't just live all day,

00:26:56

every day, we're worried about that.

00:26:57

And what can I do that's going to prolong that by a few seconds, a few hours, a few days, a few

00:27:01

weeks, a few months, a few decades, you know, just do the best that you can right now with

00:27:08

what you have to work with.

00:27:09

And AI is one of the things that you have to work with.

00:27:11

So I agree.

00:27:12

I agree.

00:27:13

All right.

00:27:13

Let's talk about four roles.

00:27:14

So we're a chapter three, four roles for co-intelligence.

00:27:17

I'll give you the four roles.

00:27:18

Uh, rule number one, always invite AI to the table.

00:27:22

Rule number two, be the human in the loop.

00:27:25

Rule number three, treat the AI like a person, but tell it what kind of person it is.

00:27:30

Uh, rule number four, assume this is the worst AI you'll ever use.

00:27:35

So Mike, I want to, before we go into like the details of this, which one of those kind

00:27:40

of jumped out to you as like the, huh, like, wow, like that's a, that's a really interesting way

00:27:46

to, to, you know, approach that.

00:27:48

Uh, it's hard for you to pick one.

00:27:51

Um, honestly, I think all of these principles are great.

00:27:55

Like this is really where I started to, okay, yeah, I, I like what you have to say about this.

00:28:02

Cause he articulates it really well.

00:28:04

And there are a couple of these things that they did jump jump out to me.

00:28:07

So number one, always invite AI to the table.

00:28:10

I have a lot of people around me who are on the other side of AI is incredible.

00:28:17

We should be using it for everything.

00:28:19

It's, um, and I've, I'm the skeptic in some of those circles, be like, well, look at the output

00:28:25

that is generating.

00:28:25

It's not that good.

00:28:26

Like, Oh, no, no, this is great.

00:28:28

You got to use this.

00:28:28

And I find myself resisting that because like the output just isn't all that great by itself.

00:28:34

So that's like, it's not really an action item, uh, but sort of like my mindset is shift around.

00:28:42

This is like, well, how can I bring AI into this process, which is how I came across that.

00:28:47

Every, or that spiral app that we talked about in the, the pro show.

00:28:51

Um, but the other one be the human in the loop, like this resonated with me because Cal Newport talked

00:28:57

back in deep work about the types of people who are going to be successful in the digital economy.

00:29:03

And one of the groups of people, there were three groups of people.

00:29:06

Um, one, one was the people who had access to resources.

00:29:10

Another one was the people who were the very best at what they do.

00:29:12

And then the third one was the one that's applicable here.

00:29:14

The ones who know how to use the machines.

00:29:17

A is a machine, right?

00:29:20

And so you got to enter interface with it the right way.

00:29:24

So a principal three, you know, treat it like a person.

00:29:26

Um, and then the, so number one, number two really jumped out at me, but then number four also,

00:29:31

assume this is the worst AI you will ever use.

00:29:33

I mean, that's really the entire point right there.

00:29:35

It's only going to get better.

00:29:37

Uh, but the more you think about that, the more, oh yeah, I guess I really should, you know, put in some

00:29:43

effort to see what this is capable of, it's not going to be wasted time because it's just going

00:29:49

to get better and the work that you put in to figure out how to write good prompts and things

00:29:55

like that, that's just going to become more effective over time.

00:29:58

Yeah.

00:29:59

So this, this worked out better than I thought it was going to because the one that jumped out

00:30:01

to me was principal three, which is tria, like human.

00:30:04

And it wasn't necessarily that part.

00:30:06

It was the tell it, what kind of person it is.

00:30:08

Yeah.

00:30:09

And like when he gives the examples throughout the rest of the book and he's like,

00:30:13

you know, answer this question like so and so or give me the perspective of so and so, you know, on blah, blah, blah.

00:30:18

I was like, Oh my gosh, I never thought to like tell it who it is because I would never do that

00:30:25

with another human, right?

00:30:26

Like I would never be like, well, sorry, I shouldn't say never.

00:30:28

I would very rarely be like Mike.

00:30:30

I want you to act like a five year old who's having a temper temperature and blah, blah, blah, blah, blah.

00:30:36

And like I would not, I would just wouldn't do that.

00:30:39

Right.

00:30:39

That's not my style, but it makes so much sense.

00:30:43

Right.

00:30:43

Like I'm interacting with this other thing that will essentially do what I tell it to.

00:30:47

If it can, or if it's not been coded some other way, oh, answer it in the following way.

00:30:53

Like, you know, respond like you're the editor of a magazine or respond like your, you know, uh, uh, uh, I don't know.

00:31:00

The CEO of some fortune 500 company.

00:31:02

And I was like, I never thought to tell it to do that.

00:31:05

Like I never would have thought to give it prompts like this.

00:31:08

So I probably would have fallen into that also, but I actually have some experience with this

00:31:14

because I've gone through like the ship 30 for 30 stuff and they've kind of really leaned into this.

00:31:17

And some of the prompts that I've used, like I have these stored in text expander because they are so long.

00:31:23

Let me just start going through one of these prompts that I have used because it's

00:31:27

exactly what you're describing, but just to give you an example of how deep this can go.

00:31:32

I'm a digital writer.

00:31:33

You're my personal idea generation consultant.

00:31:35

You are an expert in coming up with topics and sub topics that my audience will find useful.

00:31:39

I want you to help creating sub topic ideas for some of the topics I'm interested in writing about.

00:31:44

What are sub topics?

00:31:45

Sub topics are outcome focus that will help readers build a skill, implement a strategy, or solve a problem within that niche.

00:31:50

Rules when creating sub, sub topics, you must begin each sub topic with a verb to create a clear action oriented focus.

00:31:56

You must avoid using specific tactics or techniques within the sub topic as those will become my proven approaches.

00:32:01

Here are 40 potential sub topic verbs that you can use, mastering, strengthening, navigating, innovating.

00:32:06

OK, here's an example, running your first marathon sub topics.

00:32:09

And then there's a whole bunch of bullets.

00:32:11

Here's what I want to avoid.

00:32:12

Five specific things.

00:32:14

The difference between my ideal list and the list of what I want to avoid.

00:32:17

Things in here are in the what I want to avoid list are proven approaches are always there to help you get the idea.

00:32:22

So now here's the assignment.

00:32:23

You're going to help me generate seven sub topics.

00:32:25

You will ask me for the topic I want to write about.

00:32:27

When I answer, you will reply with seven sub topics and only that list of seven sub topics.

00:32:32

Do you understand if so confirmed, nasty for the topic?

00:32:34

And then what happens is that kicks off like a computer program because OK, I get it.

00:32:39

What's the sub topic?

00:32:40

OK, and then you get the sub topic and then here's the list.

00:32:42

And like I've used this and it's it's crazy.

00:32:46

Like this is this is when I started to become a believer in AI, to be honest, because I was like, oh, so it's not going to write all this stuff for me.

00:32:54

But it's going to help me ideate.

00:32:56

So I don't have to worry about what am I going to write about for the newsletter this week.

00:33:01

And you need to put a pin in that because that's a later chapter.

00:33:03

So you're not allowed to go to your print.

00:33:05

But like, I think, you know, these principles, there's nothing.

00:33:12

There's nothing unethical about these principles, right?

00:33:15

Like bring it into the thing.

00:33:17

Well, OK, what else would I bring into the?

00:33:19

Well, I'd bring in a calculator.

00:33:20

I'd bring in a textbook.

00:33:21

I'd bring in the internet.

00:33:22

I'd bring it, you know, like, I'm going to bring these other tools in here and there's no, there's no problem with that.

00:33:26

Why wouldn't I bring in the tool that is AI?

00:33:28

Then there's the be the human in the loop.

00:33:31

Well, I need to like check it and I need to verify and I need to not just trust it blindly.

00:33:36

OK, that makes total sense.

00:33:37

Like that's a good safeguard for what I'm doing.

00:33:40

Treat it like a person.

00:33:41

Tell what is you just gave a great example of how detailed you can go in that.

00:33:46

And when you do that, it kicks back such better, like, or so much better responses.

00:33:52

Exactly.

00:33:53

What I like what he says, and I think he said it before this, but it's like the AI, the chatbot, the LLM, whatever it is, it really just wants to give you what you want.

00:34:05

It wants to make you happy with its response.

00:34:07

And it doesn't have any motive behind it.

00:34:09

It's math, right?

00:34:11

Like it's math that it wants you to be pleased with the what it what it kicks back.

00:34:16

And the way it knows it's pleased with what you kick back is you're happy with the answer and, you know, kind of a thing.

00:34:22

Which I think is just awesome.

00:34:23

And then assume it's the worst that I will ever get.

00:34:25

Honestly, Mike, like I hope I hope that's where we are, right?

00:34:28

Like I hope that what we're dealing with now as good as it is is the worst that we'll have.

00:34:32

And it's just going to keep getting better and better.

00:34:34

And when I say better and better, I mean more useful.

00:34:37

Like this is the utilitarian in me coming out that says it's more useful.

00:34:40

It's more useful.

00:34:41

It's more useful.

00:34:41

It reduces my time to do the things I don't want to do and gives me more time to do the things that I want to do or it reduces the lower value activities and lets me spend more time in the high.

00:34:51

And the higher value activities.

00:34:53

And I like, I mean, I know we've we've been kind of going on this this chapter for a little while, but like I think one, two and three really lay out a good, a good path for where we're going in part two for four, four through nine.

00:35:06

Agreed.

00:35:08

And the thing that really I think speaks to me as we're going to talk about the rest of the chapters here, like how do you actually practically.

00:35:22

Safely set yourself up for success in the future.

00:35:25

Remember principle to be the human in the loop.

00:35:28

I mean, people do this already in their jobs where they just insert themselves in the processes because then the process can't happen without them and they have job security.

00:35:38

That's essentially what you're doing with the AI here.

00:35:41

OK, so now he apologizes chapter four.

00:35:46

OK, so AI is a person.

00:35:49

He spends the first part and he's like, hey, I'm going to do a thing.

00:35:51

I know it's not right, but here we are.

00:35:54

AI is a person.

00:35:55

He's like, I'm treating AI like it's another person.

00:35:57

And this is where like, I think his title of his book is so so well done, right?

00:36:03

He's going to treat it like a co-intelligence and he's really going to start talking about it like a co-intelligence.

00:36:09

So the first one, AI is a person.

00:36:12

He says, AI doesn't act like traditional software or it doesn't act like a human.

00:36:17

It excels at tasks that are that are human tasks, writing, analyzing, coding, chatting.

00:36:24

You know, basically how he talks.

00:36:29

He goes into and this is where he maybe lost me a little bit is like he goes into the Turing test and he goes into, you know, theories of mind and cyborgs and all this other stuff.

00:36:37

And I was kind of like, OK, this is this is fine, but it was a little, a little too much for me.

00:36:42

He talks about one of the things that I think is crazy scary, but these replicas.

00:36:46

So for instance, I'm going to do my best that I can to describe this, but it's like, let's say Mike dies, right?

00:36:52

And that'd be a terrible thing, but I want to keep recording, you know, bookworm and I want Mike to stay my co-hosts, right?

00:36:59

OK, so I would get a replica and I would train it on all of Mike's writing, all of his YouTube videos, all of his podcasts.

00:37:07

I would train it on everything that Mike's ever done.

00:37:10

And then I would essentially keep doing the show with Mike's replica.

00:37:14

And I'm like, this is creepy. And there if you if that ever happens, you have my permission.

00:37:20

Thanks.

00:37:22

As long as the as long as the royalty checks go to your your family, right?

00:37:26

But but it's like this is like actually one of the parts that I go, OK, we jumped over the creepy line like this.

00:37:35

I just I'm not I can't I don't like that.

00:37:38

Like I don't like it.

00:37:39

But he says that AIs will take the place of real people.

00:37:44

And that said, though, like one of the areas where I think it's actually

00:37:47

potentially very useful.

00:37:51

But it again, it gets on the border of it can be used well and it can be used poorly is say like a counselor, like they have a they have an example there where they go through, you know, this counselor helping someone out.

00:38:06

And it's like, yeah, but like what if what if the AI messes up?

00:38:11

Like what if it what if it hallucinates or what if it does a thing in a way because it's just been trained and it's just doing math and it's just trying to use weighted tokens to figure out the next word it should put.

00:38:21

There's not that.

00:38:23

Guard rail. There's not that certification.

00:38:28

There's not that training like a human would have that doesn't mean a human can who can't mess up and they and a human can't do the wrong.

00:38:36

We see that all the time that happens all across the world.

00:38:40

But at the same time, like I worry about about that.

00:38:43

So that's kind of AIs a AIs a person that's my my summary of AIs a person.

00:38:47

What do you have?

00:38:48

Well, I mean, there's a lot of stuff in here that I didn't even just capture any notes on because like the replica stuff.

00:38:55

I'm not interested in in that part of it.

00:39:00

I think the big takeaways from here, the reasons that you should view AI as a person is that it doesn't be.

00:39:06

It doesn't behave like traditional software or traditional software is predictable, reliable and it follows a strict set of rules.

00:39:14

So it seems more random like when you're working with a person.

00:39:20

So that's the justification for this approach.

00:39:23

Treat AI like a person because it often behaves like one.

00:39:27

But then the examples of the Turing test and and Tave, which was created by Microsoft in 2016 and within hours,

00:39:36

turned into a racist, sexist, hateful troll.

00:39:38

Do you remember that happening?

00:39:40

I remember hearing about it.

00:39:43

I didn't really care at the time.

00:39:45

And honestly, I remember thinking and I still think that that's just a replication of how awful people are.

00:39:55

So that's what tools do is they multiply the the effectiveness for good or bad of the the operator.

00:40:05

And I think that there there is a chance that people get their hands on this and it just multiplies the negative stuff.

00:40:13

That doesn't mean that we should avoid it altogether.

00:40:17

And if I can use it to spread the positive, then I'm going to try to to do that.

00:40:22

The one thing that that I didn't like about this chapter.

00:40:27

Well, I mean, it was an important point.

00:40:29

But the thing that kind of gave me pause is that he said that AI will soon give us each our own private echo chamber.

00:40:37

And I can see that because already people are curating their sources and they're finding facts that are going to align with their preconceived notions about what is true.

00:40:52

And because they found it on the internet, therefore, it is true.

00:40:56

And the rest of you sheep will need to wake up.

00:40:58

[laughter]

00:40:59

Like insert obligatory reference to live in all thinking here where Dave Gray talks about how the the internet is a grocery store for facts and you can find whatever you want.

00:41:10

AI is going to make that worse, especially if people are using it in a way that they're not intentionally trying to find find out what they don't know.

00:41:22

They're not intentionally trying to find what other people think other perspectives.

00:41:27

Those conversations need to happen.

00:41:30

And I feel like what AI as a person will enable people to do is go find something that sounds like a right answer.

00:41:38

They'll quickly verify it based on their own experiences and understandings.

00:41:42

It's just going to reinforce the bubble of belief.

00:41:46

And that's going to be a negative thing.

00:41:49

But what do you do about it?

00:41:51

You fight back against it wherever you can.

00:41:52

Yeah.

00:41:53

Yeah.

00:41:54

All right.

00:41:55

So we were general in chapter four.

00:41:57

That's AI as a person.

00:41:58

Now we're going to get more specific in 5, 6, 7, and 8.

00:42:02

So now AI is going to be a creative.

00:42:04

This was probably one of my favorite chapters in the book.

00:42:07

I thought the way he outlines this and the points he makes in this were very good in terms of,

00:42:14

you know, like I said, the background I come from just to give you a little bit of an example is,

00:42:19

as an educator, if I ask students to read a paper and summarize it, at this point in my career,

00:42:26

or at this point in our technology system, I don't expect that they're actually going to read it and summarize it.

00:42:31

I expect that they're going to throw it into one of these AI tools and they're going to get a summary out

00:42:35

and they're going to claim that as their summary or they're going to tweak it and they're going to modify it

00:42:39

and then they're going to say that's their summary.

00:42:41

So I have to approach this differently in terms of I would not ask, I would not have them do that assignment.

00:42:46

He talks about this, right, in terms of, you know, well, what if you had it instead generate ideas,

00:42:55

right, like, okay, so do this and summarize it from this perspective, this perspective, this perspective,

00:43:01

this perspective, and then your job is to look across all those perspectives.

00:43:05

So essentially it's the foundational seeds of creativity.

00:43:09

It's the foundational seeds of putting things together.

00:43:11

That's a bad example.

00:43:13

Like there are much better examples and he gives those, but, you know, like come up with what was the,

00:43:19

what was the example we gave Mike where it was like a wine shop or a cheese shop or something

00:43:25

and give me like a list of 20 possible names for it.

00:43:28

And, you know, like, but I forget what he was blending together.

00:43:30

And it was like, it was, it was kind of, it was very artificial, if you will, like in terms of what it was,

00:43:37

but it showed you the power of how it could do it.

00:43:39

Like it can generate a list of 100 things.

00:43:43

They might be good.

00:43:44

They might be bad, but it can easily do that.

00:43:46

You know, it can, and you gave a good example of that too with, with your example of the big,

00:43:51

the big prompt that you gave it.

00:43:53

So, you know, he talks about alternative uses tests, the remote associate,

00:43:58

associate's test.

00:43:59

These are some tests or creativity or some ways that you can get more creative.

00:44:02

I liked his, his distinction between novelty and originality.

00:44:07

Right.

00:44:08

So the way he describes this is new ideas do not come from the ether.

00:44:12

They're based on existing concepts.

00:44:13

Innovation scholars have pointed through the importance of recombination of generating ideas

00:44:19

and of obviously, you know, being Apple people, right.

00:44:22

We would go back to the, you know, what is it?

00:44:24

Well, it's not any of these individual things.

00:44:26

It's the smushing together of all three of those things.

00:44:28

And that's where the innovation really, really happened with the, with the iPhone.

00:44:31

I mean, we would, we would do this in innovation classes.

00:44:34

I would teach at Virginia Tech is instead of having students brainstorm from nothing,

00:44:39

we would say, okay, you need to put this idea, this idea, and this idea together

00:44:44

and come up with a new idea.

00:44:45

Right.

00:44:45

And that, that's what they had to brainstorm or come up with 10 new ideas around that.

00:44:50

So I really liked basically using AI as a tool to help you be more creative,

00:44:58

to help you generate more ideas, to help you in this brainstorming process.

00:45:01

I thought I just thought he did a really good job here.

00:45:03

Yeah.

00:45:05

I mean, the, he starts this chapter by talking about how the weakness of LLMs is

00:45:09

also their strength, the ability to make stuff up.

00:45:12

So the same features that make an LLM unreliable and dangerous for factual work,

00:45:20

make them useful for creative work.

00:45:22

And I think that you're right.

00:45:25

Like, I like this chapter a lot better than the previous one,

00:45:28

but I feel like the previous one is necessary for all of these other chapters

00:45:32

that he's going to be talking about here.

00:45:34

So the list of ideas, you know, the reason that that works is that AI

00:45:41

doesn't know when it hallucinates and it can't explain its answers.

00:45:45

It's not the big hallucinations, but the small ones that you don't catch,

00:45:49

that really cause the issues.

00:45:52

And I like to do this all the time, not just that big prompt that I shared,

00:45:57

but I'm working on this, this, this product for,

00:46:01

looks like an obsidian done for you vault with all the workflow stuff.

00:46:04

And I'm trying to come up with a catchy name for it.

00:46:07

There's like 19 other products that use the term life OS.

00:46:11

So that's not an option, you know, and I don't want some big, long name.

00:46:16

So I like, but once a day, I'm like, you are my, my brainstorming partner.

00:46:22

You're good at coming up with names.

00:46:24

You know, this is the, the product that I'm making.

00:46:26

Give me 25 names for this.

00:46:27

And then because you can have the conversational nature, give me 25 more.

00:46:30

Like I look at hundreds of these every single day and he points out that most

00:46:34

of the suggestions are going to be mediocre, but that's okay,

00:46:38

because that's where you come in as a human.

00:46:40

And there are ways to push AI from these average mediocre answers.

00:46:47

To the really good ones to the high variance by telling it specifically who it is

00:46:51

and what it's doing, how it can help you.

00:46:54

So I feel like there's a lot of, there's a lot of discourse right now about AI

00:47:03

and how it's going to eat all the jobs of, of creatives.

00:47:07

And I just don't believe that's true.

00:47:09

We talked a little bit about this in the pro show last time.

00:47:13

Like it really bugs me when some of the Apple podcasters that I listen to and I,

00:47:18

I like our like, yeah, we're just not going to do this.

00:47:23

And we think it's terrible that AI is going to eat these jobs of these designers.

00:47:28

Like, no, like that's, you don't understand it.

00:47:31

The good designers are the ones who are going to be able to use this and, and,

00:47:36

and it's going to enhance their workflows.

00:47:38

He talked about later on in the book.

00:47:40

So I'm kind of jumping ahead here, I guess.

00:47:42

But like a countenance in the 70s before calculators were really a thing.

00:47:47

Their job was manually doing the math.

00:47:51

And then when calculators came out, the need for accountants didn't disappear.

00:47:55

It's just their bundle of tasks was different.

00:47:58

And they used the calculators.

00:48:00

They use the spreadsheets.

00:48:02

And I think the same thing is going to happen with AI here.

00:48:05

And if you are a creator, like that, then you really need to know how this

00:48:11

is going to work because if you are a poor creator, then yeah, AI is going to eat

00:48:19

your lunch.

00:48:19

What is a poor creator?

00:48:21

Like, I don't know that I'm comfortable defining that.

00:48:23

But one of the things that he mentions, like the, the industries that's going to

00:48:26

get disrupted is like the stock art industry.

00:48:29

And we talked about that last time.

00:48:31

I am a okay with that.

00:48:32

I mean, I, I've had, I've worked with marketing agencies in the past who they

00:48:40

helped us with different things when I was working with the family business.

00:48:43

And they bought these images, but we didn't have the receipts.

00:48:48

And then years later, you know, someone from the, from Getty is like going to sue

00:48:54

us because we had this image that was used in a blog post and we didn't have the

00:48:58

license for it.

00:48:59

I don't have time to deal with this.

00:49:02

And I don't want to pay you the $10,000 that you say you're owed.

00:49:05

So you know what I'm going to do in the future?

00:49:07

I'm going to generate an AI image and get it as close to what I know I want as I can.

00:49:12

Now I'm not going to pay you a penny.

00:49:14

Like these people are just, and they're not all like that.

00:49:17

I mean, there are legitimately people who are taking good pictures and they, they

00:49:21

should be compensated for, for their work, you know, but the people who are going

00:49:24

to take the, the great pictures and they're going to create the, the great art.

00:49:27

I still think there's going to be, uh, there's still going to be a market for that

00:49:31

kind of stuff.

00:49:32

If I want a logo for a podcast that I'm developing or a product that I'm making,

00:49:39

I'm not going to ask mid journey to generate me a logo.

00:49:43

There's just so much more that goes into it.

00:49:46

There's a human element that, that, um, has to be there for that.

00:49:50

But, you know, rant over, I guess the, the takeaway here is that if you are doing

00:49:56

anything creatively, um, then you really need to figure out how to make this stuff

00:50:00

work for you if you want to be relevant 20 years from now.

00:50:03

Yeah.

00:50:03

And the last, the last point related to what you're saying, it made me think of it is

00:50:07

the most, he says the most innovative people benefit the least from AI.

00:50:11

So basically it raises the floor, you know, like I am not a good graphic designer,

00:50:16

but through using these tools, I can be a better graphic designer, right?

00:50:21

Like because when I'm starting at zero, these tools give me something.

00:50:25

But if you're a really good graphic designer, these tools aren't going to help

00:50:28

you that much because you're already really good at this, you know, like, you

00:50:31

don't need the tool.

00:50:32

That's the thing.

00:50:33

You don't, you shouldn't need to go to a professional graphic designer because

00:50:36

you need an image for an internal presentation.

00:50:39

Just use the AI, grab something good enough that communicates the idea you're

00:50:43

trying to get across and you're done.

00:50:45

The people who really are going to be producing these professional commercial

00:50:49

works, they're going to need to go work with the pros and get that stuff made.

00:50:54

And that's always going to be the case.

00:50:57

But for everybody else who just needs something quick and it doesn't have to be

00:51:01

totally professional, this, you need to know how to use this stuff.

00:51:05

Yeah, exactly.

00:51:06

Exactly.

00:51:07

Okay.

00:51:07

Let's move on to, let's move on to six.

00:51:08

So AI is a co-worker.

00:51:10

So now we've moved into AI as a co-worker.

00:51:12

I thought about you.

00:51:13

Um, as I'm working through this and it's like, you know, you had said, and I

00:51:17

can't remember where it was or where you said it, but like, you know, personal

00:51:22

assistant, right?

00:51:23

And it's like, I was like, how could Mike use AI and how is he using

00:51:27

AI and how might he use AI in the future for, you know, for, uh, for an assistant

00:51:32

to take some of the tasks that are necessary, but they're lower, they're

00:51:37

lower impact, I guess, if you will.

00:51:39

And how can you turn that over to, to, to that everyday?

00:51:42

I, before I let you answer that, uh, let's, let's do a little bit of overview

00:51:45

of what it is.

00:51:46

But they did a study on consultants and they found that AI powered consultants.

00:51:51

So the consultants that were also using AI were more effective.

00:51:54

They were more creative.

00:51:55

They wrote better, uh, more analytical than their peers.

00:51:58

Um, that there are these jobs that it will be more likely to be overtaken.

00:52:06

I guess is the best word or have overlapping portions with, with AI.

00:52:09

My job was one of those.

00:52:11

So I was like, wow, this is really interesting.

00:52:13

You know, college professor was like in the top, whatever of them.

00:52:16

And what I can tell you is like, yeah, I can totally see that that there are

00:52:20

portions of my job that the AI man, when I can give that to it and trust that it

00:52:26

does it with 99% accuracy, I'm, I'm going to do that as soon as I can.

00:52:30

Right.

00:52:31

Like I don't fear that on any level.

00:52:33

I don't think that that's a bad thing.

00:52:35

I see that as a, Oh, that's a way for me to offload the stuff that isn't the

00:52:40

higher value activity, uh, during my day.

00:52:43

Um, then tasks for us, task for AI.

00:52:46

He said, he talks about separating these tasks out.

00:52:49

Um, and like what are the tasks that I need to do and what are the tasks that I

00:52:52

could give off to the AI?

00:52:53

Then he, um, he goes into like taking that and making it into work systems.

00:52:58

What are the systems that we can then say the AI handles this part of the

00:53:02

system and the humans have to handle the, this part of the system.

00:53:05

And then he gets into more like the job side of it and it's like, okay, are

00:53:09

there actual jobs that, you know, AI might just be able to, to, to take over.

00:53:14

And none of this, like, I think what I, what I walked away happy about,

00:53:19

the happy might be the wrong word, but it's like the, the doomsday scenario of

00:53:24

like AI will take all of our jobs.

00:53:26

It was like, no, I don't think AI will take all of our jobs.

00:53:29

I think AI will redistribute tasks.

00:53:33

It'll help us develop new systems and it will change some of our jobs, but

00:53:39

like none of that was like in a bad way.

00:53:41

Like it was, it wasn't like, Oh my gosh, you know, it was much more in the way

00:53:45

of I used to have to or farmers used to have to stand behind a horse or a team

00:53:51

of horses and hold the plow as they walked.

00:53:53

And then we got machines that could do that.

00:53:55

And now I could drive the tractor, you know, like it was like, that's the way I

00:54:00

see this is what's it going to look like in the future?

00:54:02

Who knows?

00:54:03

What are the systems that are going to develop?

00:54:05

What are the jobs?

00:54:05

How are the tasks going to be distributed?

00:54:07

We don't know yet, but, but I'm more encouraged by the fact that I think this

00:54:11

is going to be a good thing.

00:54:12

Um, you know, it's not going to be good for everybody, but I think,

00:54:15

overall, in general, it's going to be a good thing.

00:54:17

I think it has the potential to be good for everybody.

00:54:20

Uh, I don't think that's, I don't think that's generalizing too much.

00:54:26

And like I said, there are certain industries that will be disrupted, mentioned

00:54:30

the stock photography industry.

00:54:31

He also mentions in this one call centers.

00:54:34

Um, but if you are working in a call center, uh, let's for the sake of argument

00:54:43

here, call that a low level job since it has the potential to be eaten by AI.

00:54:51

For as we're having this discussion, just because you are in that position, doesn't

00:54:56

mean you have to stay in that position.

00:54:58

That may be the best job that you can get right now.

00:55:00

It may be the only job that you can get right now, but I firmly believe that if

00:55:04

you work harder on yourself, then you do on your job, your skills will develop to

00:55:09

the point where you can get another better job.

00:55:13

The marketplace will reward you for the skills that you have developed.

00:55:17

Doesn't happen automatically.

00:55:18

You have to stand up for yourself.

00:55:20

You have to ask for things, but the whole idea of career capitals, Cal Newport

00:55:26

talks about it, uh, is definitely in, in play here.

00:55:28

Now, um, AI will probably affect your job as what he says at the very beginning,

00:55:33

but just because your job overlaps with AI doesn't mean your job will be replaced.

00:55:38

The danger I think we have is in the term he uses falling asleep at the wheel,

00:55:44

which is letting AI take over instead of using it as a tool.

00:55:48

He talks about throughout the book, but also it comes back here, this jagged front

00:55:54

tier, which is like there's this wall that goes all over the place and everything

00:56:00

inside the wall is stuff that AI can do.

00:56:02

Everything outside the wall is stuff that it can't do.

00:56:04

And we think that the wall is like nice and straight.

00:56:07

You know, but we find all these different points on either side of that,

00:56:12

that nice straight line that's in our mind where, Oh my gosh,

00:56:15

I can't believe that AI can actually do that.

00:56:17

Or, Oh my gosh, AI is so stupid.

00:56:19

It can't even do this.

00:56:20

That's where that perspective comes from.

00:56:22

So we got to stay, we got to stay curious.

00:56:24

And we got to constantly be trying to figure this stuff out.

00:56:27

He uses three different types of tasks here.

00:56:30

The just me tasks.

00:56:31

These are tasks in which AI is not useful.

00:56:33

It just gets in the way delegated tasks, tasks that you may assign the AI

00:56:37

and then as a human, you have to carefully check and then automated tasks,

00:56:41

tests that you just leave completely to the AI and you don't check on.

00:56:44

I think that's a good breakdown of the different types of tasks and I don't have

00:56:49

specific tasks that fall into those different categories, but I feel like that's

00:56:53

a really good setup for the centaurs and the cyborgs that he talks about,

00:56:56

the two different ways to work with AI were a centaur.

00:56:59

There's a clear line between the person and the machine, kind of like the,

00:57:02

the torso of the man and then the back of the horse, but a cyborg is like a

00:57:09

blending of a machine in person where you're working in tandem with the, with the AI.

00:57:14

So I think this understanding or this model kind of sets it up effectively for

00:57:22

thinking about how we're going to work with this AI in the future.

00:57:25

I haven't had enough time to really think about what does this mean for me

00:57:28

other than, you know, try to use more, more prompts in the creative process as we

00:57:32

talked about.

00:57:33

And this is where he, this is actually the part where it talks about the, the

00:57:37

accountants were who were in charge of calculating numbers by hand.

00:57:39

You know, your, your job is going to change and it's going to change whether

00:57:43

or not the technology that disrupts it was AI or not.

00:57:46

If it wasn't AI, there was going to be another technology that was going to

00:57:49

disrupt it.

00:57:50

My grandpa owned a type setting business and my uncle took over the business from

00:57:58

him and then when Max came out, they tried to hold on to this analog type setting

00:58:06

business, but everyone was moving to digital publishing.

00:58:08

Guess what happened?

00:58:09

You know, it's the whole innovators dilemma.

00:58:12

That was one of the books we covered way at the beginning of bookworm by Clayton

00:58:17

Christensen, who's a brilliant business business writer.

00:58:20

Was a brilliant business writer.

00:58:23

And the, the trick is not to try to protect things as they are because you're

00:58:29

comfortable, but these disruptions are going to come and you got to figure out,

00:58:33

you know, what does this mean for me and how can I leverage this AI is just

00:58:36

one of those disruptions.

00:58:38

Yeah.

00:58:39

So, so you answered my question that I was going to ask you next, which was Mike,

00:58:42

what are the tasks you're going to turn over to AI?

00:58:45

Right.

00:58:46

Like, so it got into that, you know, me delegate and then automate.

00:58:49

So you don't know that yet, but I would encourage that to be an action item for you.

00:58:52

I think that would be fun for me to hear what your answer is.

00:58:55

And then the last point I'll make here is I like the way he frames it and he

00:59:01

ties back into this idea of co-intelligence.

00:59:03

Like I really like his theme of co-intelligence here because he makes a

00:59:07

statement, how can AI be used as a co-intelligence to help us manage work?

00:59:12

And I think if we came in and like people come into it with that frame of mind that

00:59:16

says, okay, this is my job.

00:59:19

I do this thing every week or every day or twice a day or whatever it is.

00:59:24

Is there an AI that could get me 80% of the way there that way that job is so much

00:59:29

easier, better, more routine, more structured, you know, whatever it might be?

00:59:33

I think if we think about it as a co-intelligence, a partner, right?

00:59:36

Like that we're working with.

00:59:38

And I don't want to say it's a free partner because we might have to pay for some

00:59:41

subscription or whatever it is, but it's like, it's a, I don't know, it's a different

00:59:46

type of partner.

00:59:47

It's a different type of co-intelligence than you and I sharing that task, which I

00:59:52

think is a really fun way to think about this or a really like innovative way to

00:59:57

think about this.

00:59:58

And I like it.

00:59:59

That's one of the reasons why I like the co-intelligence idea.

01:00:03

Okay, so let's move to chapter seven.

01:00:06

So now AI is a tutor, obviously.

01:00:08

This is exciting to me and interesting to me.

01:00:10

I don't have a ton of like notes on this one, but this was the chapter that

01:00:15

got my, got me stewing in terms of like, what does this mean for me individually?

01:00:19

What does this mean for me personally?

01:00:20

I won't take us down.

01:00:22

Well, let me, let me hijack this one.

01:00:25

I'll share some of my notes and then I want to hear like, how are you thinking

01:00:29

this is going to impact you?

01:00:30

Okay.

01:00:31

All right.

01:00:32

So my notes from AI as a tutor.

01:00:34

And it's some, one thing that isn't in the notes, but it's important as a context.

01:00:38

You mentioned that he's a professor.

01:00:40

He actually requires his students to use AI in their assignments.

01:00:44

So while we're hearing about people who are saying, well, students might be cheating

01:00:48

because they're using AI, we have to ban the use of AI.

01:00:51

He's gone all, he's gone the other way.

01:00:53

Like, yeah, there's a possibility that they could cheat, but there's always been a

01:00:56

possibility that people could cheat.

01:00:58

And let's figure out how to, how to use this productively.

01:01:00

And I think he teaches entrepreneurship.

01:01:02

So AI is a tutor.

01:01:04

One-to-one tutoring is extremely effective in getting students to perform well.

01:01:08

And there is lots of research behind that.

01:01:11

And then the downside to that is that one-on-one tutoring is hard to do.

01:01:15

It requires a lot of time.

01:01:18

He mentions that asking AI to summarize things is taking a shortcut.

01:01:23

And that's not really what we should be doing.

01:01:25

What we want to do is we want to use this as a way to develop our skills.

01:01:33

And that doesn't mean that we're just going to increase reliance on the machines and become

01:01:38

medians because that's essentially what people thought with calculators back in the 70s.

01:01:41

In fact, a survey in the 70s found that 72% of teachers and lay people disapproved of

01:01:46

kids using calculators.

01:01:48

But then a couple of years later, it's like 90 some percent wanted to use calculators

01:01:52

in the classroom.

01:01:54

Now that doesn't mean that prompt engineering is the thing.

01:01:57

It kind of speaks against that in this chapter.

01:01:59

Prompt engineering is a useful near-term skill, but it isn't that complicated.

01:02:04

It's kind of like programming in prose.

01:02:07

He does say that AI will reduce the importance of lectures, which is probably what most people

01:02:13

think of when they think of higher education and the typical learning style.

01:02:19

So Corey, how does this affect your role as a professor?

01:02:24

Yeah, so I don't think anything he said in here was I had like a stark disagreement with

01:02:33

in terms of the way he was going.

01:02:34

It's all very sound, it was backed up by what makes sense.

01:02:38

I had actually never heard of the two-sigma problem.

01:02:40

So the two-sigma problem was actually really interesting to me.

01:02:43

I actually noted it to go try to read the bloom paper.

01:02:48

But yes, lectures bad, homework good, test good.

01:02:52

And really what this gets into is this idea of hands-on, minds-on learning.

01:02:57

And in a lecture environment, unless the person is a really good lecturer and if they're a

01:03:03

really good lecturer, they're not lecturing.

01:03:04

Like they're doing other things and they're just calling it a lecture.

01:03:08

So I teach primarily like design education and a lot of that.

01:03:13

So I mean, I don't talk often for more than 15 minutes before we're doing something.

01:03:19

We're doing a problem set.

01:03:20

We're doing an example problem.

01:03:21

You know, the students are engaging in their interaction with each other.

01:03:24

I think this is really, I think we're at the cusp of a significant change and I think

01:03:32

it'll be a leap forward in the way we do education.

01:03:36

And I think it's going to be like a big leap, like a big jump forward.

01:03:40

And I think you're going to see it divide in a couple different ways.

01:03:43

You're going to have the people who are like, nope, you know, I'm putting my head in the

01:03:46

sand and we're just going to keep doing things the way we've always done them and we're not

01:03:49

going to change.

01:03:50

Then you're going to have the people that take it too far and they're going to go full

01:03:53

AI and you're going to miss out on things, right?

01:03:56

Because I don't think you can do it with full AI.

01:03:58

I think AI can be a tutor.

01:04:00

But I think also there needs to be that human in the loop, right?

01:04:02

Like the person who is the expert in that field or is able to give the feedback and

01:04:08

give, you know, a different level of feedback is essential.

01:04:12

And then there are, I think, I think there's going to be the group in the middle and there

01:04:15

might be more groups.

01:04:16

I don't even thought about it.

01:04:17

But then there's going to be people who embrace the AI and they're going to say like, okay,

01:04:21

how do we leverage this to the best that we can to meet both the human and the in person

01:04:28

inside and the tie the technology in?

01:04:31

And I think the challenge with that middle group, which is where I would want to sit.

01:04:36

Like I want to sit in that middle group.

01:04:37

The challenge with that middle group is it changes the systems.

01:04:41

It changes the tasks and it changes the jobs of teaching and it changes and that and teachers

01:04:47

are terrible for wanting to embrace change.

01:04:51

Like they want to come in and they want to say it worked last semester.

01:04:56

I'm just going to keep doing it and the things that didn't work, I adjust those and I fix

01:04:59

those, but the things that worked, I want to keep doing that.

01:05:02

And universities, you know, like the phrases, it's like turning the Titanic, right?

01:05:06

Like, I mean, to get like significant change at a university is really, really hard because

01:05:11

there are so many different stakeholder groups.

01:05:13

That said, like I read this chapter and I think to myself, oh, there's so many opportunities

01:05:20

if we leverage this correctly and then I immediately think, yeah, but, well, let's talk about pushing

01:05:27

a boulder uphill, like, holy cow, we're pushing a boulder uphill, like to try to leverage

01:05:31

this well.

01:05:33

So I mean, that's, I hopefully I don't know if that answers your question or not, but

01:05:35

it's like, those are all of the thoughts that are stewing around in my head as I think

01:05:39

about, you know, how AI might impact me personally and then universities in general and education

01:05:45

in general.

01:05:46

Do you have any thought on who is going to end up in which camp?

01:05:51

Like, are there any themes with the people who are going to resist this versus the people

01:05:58

who are going to embrace it too much?

01:06:01

No, I mean, I don't think I can say that right now.

01:06:04

Do you mean like different disciplines or types of like faculty or teachers or what do

01:06:10

you mean?

01:06:11

Well, I guess like what I think of, I don't live in this world like you do, but I have

01:06:20

my undergraduate degree experience and then I have a Bible college degree that I went

01:06:27

back and got for funsies online.

01:06:30

And I feel like the traditional college four year degree, I went to a small liberal arts

01:06:37

school in Wisconsin.

01:06:41

I feel like that setting, I don't see it fairing well with this sort of stuff.

01:06:50

I read something on ESPN the other day about some baseball team that was a D three school,

01:06:55

it's been around for like 150 years and their school is shutting down and they made it to

01:06:59

the world series, college world series.

01:07:03

I feel like AI will make more disruption in the higher education space, which again,

01:07:14

like I don't think that's a fairly bad or good with liberal arts schools, you know,

01:07:19

40, 50 grand a year to go there.

01:07:23

You should be entitled to more as a student than to show up at the pit lectures in my

01:07:28

opinion.

01:07:30

And I just I don't know if there's a certain type of higher education that is better suited

01:07:38

to ride that way or not.

01:07:40

Yeah, I mean, I think I think you're seeing what you're saying is accurate, right?

01:07:45

Like you're seeing people make decisions when it comes to when it comes to higher education.

01:07:52

But I and I think the way I'll word it is I think AI will be a catalyst in those decisions.

01:07:58

It's going to make certain things happen faster than they were already happening.

01:08:02

So we I mean, we already get a bunch of students who come in and they're like, how do I finish

01:08:05

my degree in three years, right?

01:08:07

And we get students that come in and they're like bringing in 60, 70, 80 credits, you know,

01:08:11

and it's like, oh man, like Holy cow, like you're you're, you know, but we're still

01:08:17

in a system where for certain classes of job, that degree is the ticket in, right?

01:08:22

Like you can't even get an interview without without a degree.

01:08:25

I think those type of things are going to change if you're a motivated student and you leverage

01:08:31

AI, but then you also pair it with, you know, if we go back to the mastery book and you

01:08:36

pair it with apprenticeship and you pair it with different internship experiences and

01:08:40

these different things.

01:08:41

I mean, you're going to come out really well prepared, right?

01:08:45

The AI is going to get you.

01:08:46

It's going to help you get a lot of factual information.

01:08:49

And if you can pair that with, you know, mentorship and guidance and apprenticeships and those

01:08:54

type of things, I mean, you're going to be really well prepared to be super successful

01:08:58

in, in industry.

01:09:00

I think the, I think computer science CS, right?

01:09:04

Those are, those are going to be hit pretty hard because like the fundamental skills I

01:09:11

think are easier and easier to obtain, but the mentoring is, is where you, you know,

01:09:15

you really want the community, the design projects, learning from that person that's

01:09:18

sitting two seats down from you when you, when you can't figure that out.

01:09:21

So it's like, it's a hard question to answer.

01:09:23

I think there will be areas.

01:09:25

But what I want to do call out is he talks about that some of the people who will be the

01:09:29

best at using AI are the people with the humanities degrees because they understand the context.

01:09:35

They understand all the things that go around it.

01:09:37

And they're going to be the ones that get the most out of it because they know how to

01:09:40

leverage it the most because essentially they, they understand how to interact with people

01:09:45

and they understand how to take those interactions and combine it into the system.

01:09:48

So it's going to be a fun, I say fun, right?

01:09:51

It's going to be a crazy fun 10, 15 years as, as the stuff, you know, unfolds and, and we

01:09:57

see where, where the dust settles.

01:09:59

So it'll be exciting.

01:10:01

And that's, that's the perfect segue into the next chapter about AI as a coach.

01:10:05

Yep.

01:10:06

All right.

01:10:07

So, so we're going into coaching now and basically, you know, he gets into this idea of expertise

01:10:14

building here and that like, he talks about the 10,000 hours and really that, you know,

01:10:19

this deliberate practice versus, you know, just repetitive, rote practicing and he talks

01:10:24

about these different things.

01:10:26

But so the building expertise, you know, he would say like, you need a baseline of knowledge,

01:10:30

right?

01:10:31

So let's make sure we get a base on a knowledge.

01:10:32

Then you need to practice and you do the right type of practice, which would be deliberate

01:10:35

practice.

01:10:36

And if you put those two things together, then that's how you develop expertise.

01:10:41

And what AI can do is AI can serve as this coach that helps you figure out, you know,

01:10:48

better or more deliberate practice.

01:10:50

So if you think about this piano lessons, right?

01:10:53

Like, okay, I could go take piano lessons.

01:10:57

And if I have a good coach, that coach is going to push me and push me and push me, but

01:11:01

they're going to know the right way to push me.

01:11:03

They're going to know how to push me and say like, Oh, your fingering on your, on your

01:11:07

right hand isn't good.

01:11:09

So what we're going to do is we're going to do like the next three weeks is going to

01:11:12

work on fingering on your right hand.

01:11:13

And I'm going to hate every minute of it, but I'm going to get significantly better.

01:11:17

So does AI have the ability to do that?

01:11:19

Right?

01:11:20

Like it has the ability to identify these areas where you're not as good in this particular

01:11:25

area and help coach you to that.

01:11:28

I think if I'm getting it from remembering correctly, this is where he talks about the

01:11:32

doctors, right?

01:11:33

And how the apprenticeship and like the intern level of the doctors might start going away

01:11:40

because the doctors can use AI.

01:11:43

So therefore they don't need the lower level interns to do some of that stuff for the

01:11:47

them.

01:11:48

And they just keep doing what they're doing and then use the AI for that.

01:11:50

But then we're not going to train expert doctors anymore because they're missing that

01:11:54

middle phase.

01:11:55

So it's like the removal of this middle phase is a, I remember being a big, big part of this

01:11:59

chapter.

01:12:00

Yeah, they're using the surgery robots as well, which means that the, the apprentices

01:12:06

in the room, uh, don't get the experience that they would have previously.

01:12:11

I think that's the big takeaway for me from this is that, um, yeah, you can use AI as

01:12:16

a coach and you can help it to, uh, give you feet, like you can use it to get more feedback

01:12:22

loops and, and, uh, do deliberate practice.

01:12:26

But the, the dangerous side of this, I think is that the experts become experts through

01:12:32

that deliberate practice.

01:12:34

And because their time is valuable, they're, if they just go all in with AI, there's a,

01:12:40

a major apprenticeship gap that's going to develop where the next tier is not going to

01:12:46

have an opportunity to develop the same types of skills.

01:12:51

Um, I, I think that, uh, it's probably not, uh, that sensational.

01:12:58

I think that's just kind of like a red flag and the, the practical application of this

01:13:02

is if you're someone who has those, those skills, it's worth investing in, uh, somebody

01:13:07

else, but the big question that you're asking regardless of whether you are an expert or

01:13:12

not is where is the appropriate place to insert AI as a co-intelligence into, into, uh, what

01:13:19

I do.

01:13:20

Um, he does mention in this chapter that humans working with AI co-intelligence always outperform

01:13:26

the best humans alone.

01:13:28

So again, it's not something that, oh, this is bad and it's going to eat our lunch and

01:13:32

we need to avoid it.

01:13:33

If you really want to be the best at what you do, you have to, uh, you have to embrace this

01:13:38

stuff.

01:13:39

Um, but also I think for me, one of the things that I'm always cognizant of is the type of

01:13:46

legacy that I'm going to leave and I want to help other people do more.

01:13:52

So as it pertains to me and how I would apply this, it's, you know, use the AI where appropriate,

01:13:59

but don't just neglect the people, uh, for the sake of the productivity.

01:14:04

That was something I picked up from Chris Bailey's book, The Productivity Project way

01:14:08

back in the day.

01:14:09

Uh, because he mentions that people are the reason for productivity.

01:14:12

A view AI is like a productivity enhancement tool, um, but the people are still the, the

01:14:18

most important thing.

01:14:20

And I've aired on, on the side of, well, I just got to get the task done before, you

01:14:24

know, and damage the, the relationship.

01:14:26

So I think AI just, you know, provides more opportunities for that sort of thing to happen

01:14:30

if we're not careful, but doesn't mean that it necessarily has to.

01:14:34

Uh, I don't think the fact that AI makes things more efficient means that people are

01:14:39

just going to go inward and they're not going to train up other people, uh, to, to do things.

01:14:45

I think that's kind of inherent in us is that we want the, the people that we are investing

01:14:51

into, whether those are our natural kids or, um, just like your, your students, you know,

01:14:56

you want them to go leave a bigger dent in the universe.

01:15:00

That's actually the, my life theme is I'm, I, I condense it down and I say I'm a multiplier

01:15:06

because I want to help people multiply their time and tell and leave a bigger dent in the

01:15:09

universe.

01:15:10

That's the most rewarding thing for me.

01:15:12

If I can help you do that, you know, then I'm going to do that.

01:15:15

And if I'm constantly focused on just like, how do I get more widgets cranked?

01:15:19

Then I'm going to miss those opportunities.

01:15:23

I think this is the area where this agreement might be, um, the wrong word, but I'm less

01:15:29

optimistic.

01:15:30

I'm more skeptical in this area where I think AI right now is really good at that baseline

01:15:36

knowledge.

01:15:37

It's really good at helping me get that, that baseline knowledge.

01:15:39

It's really helped good to help me edit things and, you know, answer questions and summarize

01:15:44

and do some of those things.

01:15:46

But when it comes to the coaching side, I still think like an actual human sitting across

01:15:51

from me face to face or even virtually just is so much more effective and efficient.

01:15:56

Um, because they can read cues better.

01:15:59

They can hear the nuances or understand the nuances.

01:16:03

They're less likely to be manipulated.

01:16:05

Um, you know, as you're working through things.

01:16:07

So this is probably the area where I'm a little less, you know, gung ho on what, um,

01:16:12

Moloch saying in this section.

01:16:14

All right, let's, let's get the rounding it out.

01:16:16

So now AI is our future.

01:16:17

He talks about four possible futures here.

01:16:20

He basically says AI, this is good.

01:16:21

It's going to get, it's not going to get any better.

01:16:23

Um, there's some reasons for that.

01:16:24

He throws it out there that basically, um, rules and regulations, stop it.

01:16:29

Technology progress stops it, you know, whatever it is, but it just doesn't get any better

01:16:33

than it is right now.

01:16:34

He talks about slow growth.

01:16:36

The fact that, you know, not everything can remain exponential and maybe just will be limited

01:16:40

by the future of the way the technology pans out.

01:16:43

So therefore it'll grow, but it's not going to grow as quickly as it has, you know, lately

01:16:47

or, um, from the, from the start.

01:16:50

He talks about exponential growth and he uses your, um, your flywheel, right?

01:16:54

So yeah, he talks about the fact that it's a flywheel.

01:16:56

So, uh, basically AI companies can use AI systems to help them generate the next version

01:17:01

of AI and boom, boom, boom, boom.

01:17:03

And it just feeds on itself and we, we keep that curve going up exponentially.

01:17:06

And then he talks about the machine God as the, as the last one, what they basically

01:17:11

says, we're going to develop some artificial super intelligence.

01:17:14

It's essentially going to rule everything, um, in, you know, that that's how it goes.

01:17:20

Or that's how it would go there.

01:17:22

Uh, he, I'm trying to remember what he, he thinks he doesn't think that as good as it's

01:17:26

going to get, um, is, is likely.

01:17:29

He also doesn't think the machine God scenario is the likely scenario, but I can't remember.

01:17:34

Do you remember, whether he thinks it's going to be slow growth or exponential?

01:17:37

Does he ever say that?

01:17:39

I don't remember.

01:17:40

Um, if I had to pick one, I think it's probably between slow and exponential.

01:17:46

I feel like we've seen some exponential growth recently, but I think we are going to hit

01:17:51

some diminishing returns with some of the AI stuff soon.

01:17:56

That doesn't mean that it's not going to continue to improve.

01:18:00

But I think it is going to slow a little bit from where it is right now.

01:18:03

However, I think just the rapid pace of, of change, um, that is accelerating overall.

01:18:11

So I don't expect it to be, uh, slow growth either.

01:18:15

Uh, I want to real briefly mention the, the machine God scenario.

01:18:19

He mentions that there's no theoretical reason this can't happen, but there's no reason to

01:18:23

suggest it might, which I think is kind of refreshing that, that perspective, because,

01:18:32

uh, the stuff that I've seen online from people who fall into, you know, this is as good as

01:18:36

it gets or the other extreme, it's going to take over everything.

01:18:40

They tend to be very cut and dry.

01:18:43

Like this is the reason and this is, you know, factual, 100% certainty.

01:18:47

And I feel like, uh, he's essentially saying it, at, at this chapter, we don't know what

01:18:53

this is going to look like.

01:18:55

He's basically throwing out a whole bunch of different, um, possibilities.

01:19:01

And, uh, I think I really, I, well, I really like this, uh, this approach.

01:19:05

And I think what this does for me after I read this is question anybody who says with

01:19:11

any degree of certainty, like this is what you got to watch out for.

01:19:15

This is what's going to happen.

01:19:17

Yep.

01:19:18

Uh, I think the things he says might happen where people get left behind, people struggle

01:19:24

with the, the transition.

01:19:25

I think all those, those things make sense, right?

01:19:27

He calls them the smaller catastrophes.

01:19:29

I think all those make sense, but, um, I like what you said in terms of, you know, there's

01:19:35

no probability or there's, sorry, there's no likelihood that it, it will happen, but

01:19:39

it might, you know, like we can't say, we can't say that it won't.

01:19:42

All right.

01:19:43

Um, now we have the epilogue.

01:19:44

Um, I, I didn't get a ton out of the epilogue.

01:19:47

Um, basically it's a tie back into let's think about AI as a co-intelligence.

01:19:54

Um, it doesn't have a mind of its own.

01:19:57

Its math and it's, you know, programmed and it's, it's really impressive for what it can

01:20:02

do.

01:20:03

Uh, but you know, his last statement, I think, uh, or one of the last statements he says,

01:20:07

"humans are far from obsolete at least for now."

01:20:09

Right?

01:20:10

And I was like, what a, but, but even there, like I thought it was a great way to end the

01:20:13

book.

01:20:14

It's like, hey, remember, use this thing as a co-intelligence, but humans are so important

01:20:19

and, and don't lose the sight of the fact that humans are important.

01:20:22

Yeah.

01:20:23

But he also said a, this is only like two pages and half of one of the pages is the return

01:20:29

of the prompt that he, but one of the things he mentions on page two, 12, the very last

01:20:35

page is that AI is a mirror reflecting back at us our best and worst qualities.

01:20:41

I like that too.

01:20:42

And then I also agree that that last sentence of the book is, is great about humans being

01:20:46

far from obsolete, at least for now.

01:20:48

Um, that, that's the perfect way to, to end it.

01:20:50

I agree.

01:20:51

Don't worry, but this is going to change.

01:20:54

All right.

01:20:55

So before we go to styling rating, action items, all that stuff, I have a thing for you, Mike.

01:21:01

So if you scroll down to the bottom of the notion page, I asked you not to look at these.

01:21:04

Hopefully you, you helped turn to that.

01:21:06

Okay.

01:21:07

So go into the first one, go into a four-quary, like the person.

01:21:10

All right.

01:21:11

This is the prompt I gave it.

01:21:12

And I base, I mean, we were doing a book on AI, right?

01:21:14

I had to, I had to go and see what it did.

01:21:18

Act like you're a professional book reviewer and provide a chapter by chapter summary of

01:21:21

the book, Co-intelligence Living and Working with AI by Ethan Mullick.

01:21:24

Um, I don't want to go through all these because it actually does, it does this and

01:21:28

it does it very, very well in terms of giving us a, a summary of it.

01:21:32

But now go back and go to the second one.

01:21:35

Okay.

01:21:36

All right.

01:21:37

Got it.

01:21:38

Now I said, act like you're a professional book reviewer, provide a chapter by chapter

01:21:40

summary, Co-intelligence by Ethan Mullick.

01:21:42

You should include all headings and subheadings contained within the chapters.

01:21:46

And I think it did a really good job.

01:21:49

Right?

01:21:50

Yeah.

01:21:51

I actually thought it was going to kick back an air to me that said, I can't do that.

01:21:55

It's copyrighted work.

01:21:56

You know what I mean?

01:21:57

Like, or, you know, you're going to ask me for too much of the, of the copyrighted work.

01:21:59

And it did not do that.

01:22:01

It 100% gave us a, um, a good summary.

01:22:05

Um, but what I want to call out here is one, the prompting.

01:22:11

So we've talked about this a little bit before, like the more detailed the prompts you give

01:22:14

it, the better results you get.

01:22:17

Two, you have to review this stuff to make sure that these are actually the headings

01:22:21

and the subheadings because it might have just made this stuff up.

01:22:24

Like if I don't review that, but then three, and this is not for you and I to claim, you

01:22:29

know, like the importance of ourselves or anything like that.

01:22:33

There's a big difference between ask me asking it for a summary with headings and subheadings

01:22:37

and then listening to two people talk about chapter by chapter, you know, the book and

01:22:43

what they got out of it and how they went.

01:22:45

And I say that because it's like it's, there's a place for all of these things and you can

01:22:51

use them in the way that makes the most sense for you and in a way that provides the most

01:22:55

value for you.

01:22:56

And I just think that's a, that's a valuable thing.

01:22:58

Like I really think, you know, that having the access to this is better than not having

01:23:04

the access to it.

01:23:05

100%.

01:23:06

Awesome.

01:23:07

All right.

01:23:08

Let's go into action items.

01:23:09

Mike, what action items do you have coming out of reading co-intelligent?

01:23:17

I just have the one that you gave me figure out what I'm going to delegate to AI.

01:23:23

I think I'm going to take other action from this book, but I don't have anything specific

01:23:29

that I'm going to do with it.

01:23:31

I think the big thing is to consider where I can invite AI to the table.

01:23:39

But I don't know exactly where that is yet.

01:23:42

So we mentioned the Pro Show, some of the tools that we're using that have AI.

01:23:49

But I want to find some additional ways to do some of the things that I'm doing.

01:23:54

And my big perspective change walking away from this book is essentially even if it doesn't

01:24:00

get it right on the first try and it takes a little bit of time to craft the prompt or

01:24:05

to dial it in.

01:24:06

It's worth the work to do that kind of stuff.

01:24:10

So I don't know exactly what that's going to look like, but I just want to have that

01:24:15

perspective as I go about my creative routines and build my systems.

01:24:20

Wonderful.

01:24:22

So I have two.

01:24:24

So one, I want to plan how I can teach with AI more intentionally.

01:24:27

So I am very much an adopter of this.

01:24:29

I don't fear this.

01:24:31

I welcome this into my classrooms.

01:24:33

I've already used it in certain classes.

01:24:36

And I want to be even more intentional in the 24/25 academic year about we're going to

01:24:42

embrace this.

01:24:43

We're not going to fear it.

01:24:44

We're not going to run away from it.

01:24:47

And you might cheat, sure.

01:24:50

But at the same time, I need to be creative about the way I design prompts and assignments

01:24:54

and reflections to where it's hard for you to cheat.

01:24:58

You can use AI, but we're going to do it in a way that it's difficult for you to cheat

01:25:02

easily.

01:25:04

The second one is I want to think about the scholarly side of my work.

01:25:08

So the research side, the other thing I do that isn't just in the classroom and say,

01:25:13

how do I utilize AI more effectively to do reviews or to analyze different sets of data

01:25:24

or what it might be?

01:25:26

Data is hard because if you're under IRB review, you can't really share the data because it's

01:25:33

like you're putting it out there for essentially everybody.

01:25:36

So it's protected data.

01:25:38

But more generally, secondary data.

01:25:40

How could I analyze secondary data and say, hey, go to the IPEDS database and tell me

01:25:46

the number of engineering graduates since 1974?

01:25:50

It's like, can I manipulate AI in a way that it will actually kick back an accurate response?

01:25:56

I want to play around with more in-depth prompting and utilizing it more.

01:26:00

I don't think that has a real tangible next week.

01:26:05

I say, yeah, I did exactly that.

01:26:07

But I think it's one of those ones that's in the back of my head that I'll keep thinking

01:26:10

about.

01:26:11

Awesome.

01:26:12

All right.

01:26:13

So this is my book.

01:26:14

My book means I get to rate it first.

01:26:18

I don't think you'll be surprised to hear that I loved this book.

01:26:22

I thought this book was outstanding.

01:26:26

I didn't think the author wasted much time.

01:26:29

I thought the author talked about things in a flow that made sense and organized it effectively.

01:26:36

I thought that the author hit--

01:26:39

I can't think of something that I would have wanted to be in the book that wasn't in there.

01:26:44

That it's like, oh, yeah, but you didn't talk about this whole area or this whole section.

01:26:49

I thought it was both theoretical in terms of telling us some of the fundamental understandings

01:26:54

of it, but also practical in pushing us to be in a co-intelligence relationship with

01:27:02

AI.

01:27:03

Overall, loved it.

01:27:05

I'm a huge fan of this book.

01:27:06

I will recommend it to people to read if they want to get more into understanding AI and

01:27:13

how AI we can interact with AI.

01:27:15

So you won't be surprised.

01:27:17

It's a 5.0 for me.

01:27:19

All right.

01:27:21

Well, I agree that this is probably best-case scenario for me in terms of a book on AI.

01:27:28

I enjoyed this way more than I thought I would.

01:27:32

I think the only thing that's going to keep me from rating it 5 stars is that I don't

01:27:39

know that this is going to hold up real well just because AI is changing so fast.

01:27:50

Everything he shares here makes sense.

01:27:53

I really like the four principles.

01:27:56

I really like the four scenarios, but I feel like in 12 months, does one of those scenarios

01:28:04

feel like, oh, that's just ridiculous?

01:28:07

I don't know.

01:28:08

I'm not sure that this stands the test of time.

01:28:11

I think right now, this is the book to read on artificial intelligence.

01:28:18

This is episode 200 of Bookworm, though, and we do an episode every other week.

01:28:25

So I've been reading books and talking about them for what is that?

01:28:31

Like almost eight years.

01:28:32

Eight years.

01:28:34

When we get to episode 400, is this one of the ones that I look back and like, yeah,

01:28:40

that was life-changing?

01:28:42

I'm not sure.

01:28:43

I'm just like, it might be, but I'm a little bit skeptical.

01:28:48

So I'm going to rate it four stars only because I've got a whole bunch of books that I've

01:28:54

read and a lot of books that I have rated five stars that, you know, when I read them,

01:29:00

I really felt like this book is going to revolutionize my life.

01:29:05

I'm going to be able to look back and say, I thought one way prior to this.

01:29:09

Now I think differently and my life is different as a result.

01:29:14

I think jury's still out with the artificial intelligence stuff.

01:29:18

Not that it's not going to affect our future.

01:29:22

I think it absolutely is going to change everything.

01:29:28

I think at the moment, you know, this is the best book that I've come across on the topic,

01:29:33

but like I said, it's changing so fast.

01:29:35

I'm not sure how good it holds up.

01:29:39

But I really enjoyed it.

01:29:40

And you know, I want to reiterate, you know, if you're looking for a book on artificial

01:29:45

intelligence, this is the one to read.

01:29:48

I feel like it does a really good job of explaining the different scenarios.

01:29:53

So you're going into things with your eyes open in terms of the possibilities.

01:29:58

He doesn't shy away from some of the bad stuff that could happen from it, but that's not

01:30:02

a reason to avoid it.

01:30:03

Like it's a very balanced approach, I feel.

01:30:06

And it's the right approach for where we are right now.

01:30:09

I'd agree.

01:30:10

All right, Mike, let's put co-intelligence on the shelf.

01:30:16

And let's think about what our upcoming books are.

01:30:19

So you've got the next book.

01:30:20

What's next?

01:30:21

I do.

01:30:22

The next one is Right Thing Right Now by Ryan Holiday.

01:30:25

I think we've read just about every Ryan Holiday book for Bookworm.

01:30:30

This is the third one in the Stoic series.

01:30:32

I am getting more and more excited to talk about this one.

01:30:36

I heard Ryan share that he thought this one was going to be the hardest one to sell because

01:30:42

he packaged the four Stoic virtues.

01:30:46

And he sold that idea to his publisher.

01:30:48

And he told them, hey, just so you know, this one's going to do the worst out of all of

01:30:51

them.

01:30:52

But it was the quickest one to show up on the New York Times bestsellers list.

01:30:57

So.

01:30:58

Awesome.

01:30:59

All right.

01:31:00

So then after that, as you're prepping, we'll do a book.

01:31:03

My next book is called Uncoming Greatness by Mark Miller.

01:31:07

It's a book on leadership.

01:31:08

And I found it, you know, looking through some different things on thinking about leadership.

01:31:13

And I think it'll be a good, a good book following Right Thing Right Now.

01:31:17

Mike, do you have any gap books right now?

01:31:21

Uh, maybe.

01:31:24

I do have a book with me, which I want to read.

01:31:27

I think I mentioned this as a gap book before, but didn't get a chance to read it.

01:31:30

And that is Simple Marketing for Smart People by Billy Boris and Tiago Forte.

01:31:36

Billy is the guy that does the marketing for Tiago and he has his own courses that he,

01:31:42

he sells.

01:31:44

I think he calls it like the five light bulbs method or something like that.

01:31:49

So I've seen some of his stuff before and it really does feel like very helpful, practical

01:31:55

marketing advice and as I'm getting ready to release a new product.

01:31:59

This is something that I'm thinking about.

01:32:01

Alrighty.

01:32:02

Uh, I don't have any gap books.

01:32:04

I don't want any gap books.

01:32:06

I am not in a place right now where a gap book makes sense.

01:32:10

I'm going to read Right Thing Right Now.

01:32:12

I'm going to enjoy Right Thing Right Now.

01:32:15

If I get lucky and am able to carve out some other time for a gap book, I will do that.

01:32:19

But as of right now, I do not have one.

01:32:22

Alrighty.

01:32:23

Alright.

01:32:24

So that, uh, that does it for another episode of Bookworm.

01:32:27

Uh, thanks for everybody for listening.

01:32:29

For the pro members, thank you.

01:32:31

If you're interested in the pro show, uh, we have a Patreon.

01:32:36

So you can go to patreon.com/bookwormfm.

01:32:38

Uh, there at $7 a month, you can sign up.

01:32:41

Um, wanted to support the show to get, um, bonus content in terms of the pro show.

01:32:46

Today we talked about the AI that we're using and like the apps that have it integrated

01:32:52

and how it integrates.

01:32:53

Um, you also get the, um, the bootleg version of the feed.

01:32:57

So straight from our Zoom recording, uh, we pump it out and, uh, give it to you.

01:33:01

Um, and then, um, there's an ad-free episode of the show.

01:33:05

So, um, if you're interested in that, uh, we'd love it.

01:33:08

It's, uh, patreon.com/bookwormfm.

01:33:10

Um, other than that, any last words?

01:33:14

Mike?

01:33:15

Ah, if you're reading along with us, pick up Right Thing Right Now by Ryan Holiday.

01:33:19

And we'll talk to you in a couple of weeks.