If I had to use one word to describe

modern digital tools, it would be fast.

Now, whether we're talking about

workplace communication or swarms of AI

agents or even apps for ordering food,

the focus always seems to be on reducing

friction and increasing the options

available to get people through whatever

task they're thinking about as quickly

as possible. Now, the speed of course

has a a negative side effect. It's

exhausting. It reduces so much of our

life to a a frantic blur of swipes and

taps and clicks all in a sort of

never-ending battle to keep up with a

ceaseless barrage of incoming

information. But what can we do? Going

faster has to make us more productive,

right? Or does it? You see, there's a

growing subculture of individuals who

are embracing simpler technologies that

offer less features and more friction.

And they're not doing this as a

political statement. And they're not

doing this because they're nostalgic,

but instead because they think these

more minimal tools actually makes them

better at their work and makes their

life more livable. I call this movement

slow technology, and it's what I want to

talk about today. Now, to help us better

understand it, we're going to be joined

by Amy Timberlake, who's an acclaimed

best-selling author of children's and

middle-grades books. Timberlake has won

countless distinctions, including a

Newberry Honor and an Edgar Award. Her

titles have been named to, as far as I

can tell, just about every best book of

the year list in existence. So, she

knows what she's talking about. Now,

Timberlake came to my attention

recently. Here's why I want to talk to

her about slow productivity. She came to

my attention when I discovered she had

recently shifted to using a mechanical

vintage typewriter

for more and more of her writing and

revision process. So, I really wanted to

find out what was going on here. So in

our conversation uh we talk about what

is it succeed what does it take to

succeed in that world of children's book

writing the details of her creative

process and how it's evolved and how she

came to believe that using a mechanical

typewriter was actually going to make

her a better writer and a happier human.

Spoiler alert I kind of end this

conversation about half serious about

buying a typewriter myself. So beware

about that. Um, and then after our

conversation, I'm going to step back and

I'm going to isolate some general

principles of slow technology. I'm going

to show you some other examples of slow

technology that have become popular in

recent years and make the argument that

many more people, not just those who are

in creative fields like writing, should

consider embracing simpler tools. So, if

you are tired of being told by, you

know, tech leaders like Sam Alman that

your future has to be orchestrating

armies of hyperactive AI agents, or if

you find your smartphone to be an

overstimulating anxiety machine, then

this episode is for you. As always, I'm

Pal Newport and this is Deep Questions,

the show for people seeking depth in a

distracted world. Today's question,

should I embrace slow technology? and

we'll explore some answers right after

the music.

All right, Amy, uh, thanks for joining

us today. We have so much geeking out on

creative process and technology and the

deep work to get into today. So, um, I

have been excited about this. Uh, I want

to start by just making sure that the

listeners understand your story and

where you're coming from. Um, first of

all, how do you describe even the genre

of books you write in? I I see them

described as children's book, but

there's there's a range here, right?

From, you know, books that I might think

of as like middle grades to illustrated

books. So, just as let's just set the

stage. How would you describe yourself

as a writer?

>> Uh, well, I would describe myself as a

writer who will write pretty much

anything. However, I have uh been

published in writing for kids and I've

written picture books.

Actually, I published one picture books

and then I've written middle-grade

novels, which is kind of the middle

range um of readers. And then YA would

be high would be older. So, I write

pretty much middle-rade novel age. And I

would say

that's

actually at this point it's probably

second grade through

8ighth grade, 7th grade. Um, and this

last and and this last group of novels

that I wrote, they are actually

readalouds. And the idea was that I was

trying to write something that somebody

could read in a room and everyone from

the 80 year olds to the five-year-old

would enjoy. And this was a big

challenge. I always take a challenge

every single time I take a project. And

uh yeah, and also I was trying to write

humor for the first time. I'm a quirky

writer, but I wanted to just try to be

funny in books. And

that was a big that was a big challenge

and and fun and I'm glad I tried it. I'm

really glad I tried it. So that's these

latest ones. The latest ones are the

skunken badger book. So skunken badger,

egg marks the spot and rock paper

incizers.

>> I mean so something I've always wondered

about that genre is that it's a genre

where people often they have this very

naive view of it. Right? When people

think about books they often think oh

the hard part about writing is like the

quantity of words. So like okay I get

it. I'm not going to be a big novelist

because I can't imagine writing that

many words. But then they think of a if

they're looking at like a picture book

or something, they're like, well, the

actual quantity of words there is not

that much. I could sit down and write

that many words and it would sound

roughly like a story. I could do that,

you know, like this afternoon.

>> It's a very competitive market, though.

>> What What did you discover that you then

put into that uh that first book that

like makes projects sell in children's

books because there's so many

submissions in that world? because so

many people try. What is it that people

don't understand about what a successful

book in that genre must do?

>> I think it just

well I can only speak for my experience

but I would say

the better written it is the the more

chance you have. And that first book,

I wrote it like my grandfather telling

the story. So that voice was very

specific. New Mexico

tall tale. And I just went for it. I

didn't I didn't I didn't do a lot of

things that

they might that you might read about and

hear that you have to do. I didn't um I

didn't worry about language. I didn't

worry about vocabulary. I worried about

how well can I tell this particular

story

um with this particular voice and I just

gave it everything I had. Um

so that's that's how I do it. Um it

means it means for me though that as a

writer I I'm not quick at what I do. It

means I produce a lot of words and then

I cut down. In particular for these

skunk and badger books,

they are really tightly written. They're

almost like a farce

and they can't carry a lot of extraneous

words. So I have to really know the

characters. I have to write pages and

pages and pages. And then I cut cut cut

cut cut cut cut cut cut cut cut cut cut

cut cut cut cut cut cut cut cut cut cut

cut. and the cutting just goes on

forever.

>> Yeah.

>> Um, so it's so for me it's and it's it

is really all about language and how

does the words sound and so you know I'm

often like when I write these I write

long then I stand up and I read them out

loud so that I can hear how the language

is hitting on the page and then I I cut

them again. So yeah. So I I can't really

honestly tell you what makes it

successful. I do think that the thing

that I I really do is I I I put 100% of

my effort in it.

>> Yeah.

>> So

>> I think you did just tell us. I mean

what what I'm hearing there is something

I've heard before which is in that type

of writing it's it's like a puzzle

coming together. the every word matters,

the the tonality, the way that the the

the rhythm and poetry of how it sounds

out loud. Nothing wasted,

>> everything moves it forward, which is

different than, I don't know, if I'm

Neil Stevenson writing a thousandpage,

whatever. Not every sentence needs to be

>> perfect, right? I'm like all over the

place, postmodern degression going on,

and it's fine. It's like I'm moving like

a plot. Yeah. But

>> yeah,

>> uh I've heard the same thing about

screenplays that people as with

children's books, people like, "Oh, I

could write a screenplay because I know

movies and if I look at a scene, I was

like, they're just talking. I can just

talk." And then if you talk to

professional screenwriters, like I've

interviewed them for the show, they're

like, "Oh, no, no, there's it's like a

jewelbox puzzle. Nothing can be wasted.

You can't have a single person say a

single thing that doesn't have a reason

why they're saying it. And it better

not, you know, it's got to be check

off's gone. If you mention this here,

this better come back here." And

actually, it's the nothing's wasted. So,

I I get that the sort of

>> um every sentence has to be right. But

that's interesting. You write a lot

>> so that you can then pull out as opposed

to

>> building the smaller number of sentences

very carefully just from scratch. You

you so you pull back to it. And is that

because it

>> it puts you in the having the larger

amount of text really puts you into the

moment in the character so that you can

then better find

>> the right line for that page. age. Is

that a way to think about it? Hey, let's

take a quick break to hear from some of

our sponsors.

Hey men, you need to start taking better

care of your skin. What's the best way

to do this? With Calira Lab. Calira Lab

makes high performance skin care

designed specifically for men's skin.

See, men's skin is 25% thicker, oiler,

and aged differently from women's, which

means men need clean formulas engineered

for their biology. Now, Caldera Lab's

three-step regimen is powered by clean,

clinically tested ingredients and

breakthrough patent pending technologies

that deliver visible results. Step one

is to use their cleanser, which will

clear dirt, oil, and sweat that is built

up on your skin. Then, step two, you use

their serum, which is clinically proven

to reduce wrinkles, firm skin, and

improve elasticity. It also can deal

with puffiness under your eyes. Trust

me, man, as you get to a certain age,

that becomes relevant. And step three is

use their moisture mo moisturizer which

is lightweight and non- greasy. Now I

really hadn't done much with my skin

until I got these three products from

Calerolab and I didn't realize how much

I had been missing. So this is a small

habit that produces big results. Go to

calderab.com/deep

and use the code deep to get 20% off

your first order. This episode is also

sponsored by Better Help. So, it's tax

season, which means finances are on our

minds. And this matters for our mental

health because financial stress can take

a serious toll on your mental health.

Money worries can bring anxiety and

sleep disruption and depression. It's

also probably one of the leading sources

of conflict among couples. Now, given

that 88% of Americans are feeling at

least some sort of financial stress,

according to an early 2026 survey, this

is a problem that we cannot ignore. One

part of the solution, of course, would

be getting better financial advice. But

another part might be getting the proper

support for your mental health. And this

is where Better Help enters the scene.

If you're considering therapy,

BetterHelp makes it easy. With over

30,000 therapists, BetterHelp is the

world's largest online therapy platform,

having served over 6 million people

globally. And it works with an average

rating of 4.9 out of five for a live

session based on over 1.7 million client

reviews.

So, when life feels overwhelming,

therapy can help. Sign up and get 10%

off at betterhelp.com/deepquests.

That's betterhp.com/deepquests.

All right, let's get back to the show.

>> Yeah. So, I think I think the reason why

I go long is for a couple of reasons.

Sometimes it's because I need to

discover who the character is, and the

characters are always very specific, you

know? So,

I don't know. Have you have you seen

Sally Wayright's stuff on BBC like Riot

Women or, you know, any of that?

>> I'm familiar with it. Not I I know the

the titles but I'm not very familiar

with it. She is just she's just exactly

to me exactly the kind of writer that

you were just talking about which is

where every

every piece of dialogue that she has in

one of those those um miniseries BBC

shows um is always just thought through

and so and I would imagine that she also

writes long. she has to figure out who

those characters are exactly and and

sometimes she's going back and placing

dialogue early because it makes sense

because this, you know, it's all just

figuring out where things land. Um, but

I do it I do it because it's a lot about

characters and finding those scenes. And

then also sometimes if you write long

and you really love language like I do,

you find that sentence that is just it's

perfect. It's dynamite. And you go,

okay, that's it. That's the voice.

That's what I need. Now I have to write

this whole thing again because I have to

have that kind of like whatever that is.

It's like a music. It's like a sound and

you get really excited about it.

>> Yeah.

>> Um so anyway, so that's why I do it.

>> How does that change when you're writing

the slightly older grades books that are

chapter format, more text? Um how does

that adapt to that?

>> Oh, that's that's when I'm really doing

it.

>> Interesting. I mean, yeah. All of these

all of these skunk and badger books are

really like Okay. All right. Here's

Skunk and Badger. The first line is of

this one is, "The first time Badger saw

skunk, he thought puny and shut the

front door." Took me a while to find out

that he would use the word puny and that

that is the first time, you know, that

he he shut the front door, but that's

the beginning. And so I just had to I

had to write right write until I found

that thing and I was like all right

that's how it is. That's that's what

this is.

>> Had you written like dialogue a lot of

dialogue and then sort of found in that

dialogue, oh this is the this voice

feels right for the character and then

you knew how to go back and write that

first word in that first line.

>> Yeah. Yeah. I had to find out and I had

to find out who Badger was. And uh

Badger is a is a important rock

scientist. He does important rock work

and he's always actually it's funny it's

funny that I'm talking to you about this

because Badger Badger always kind of the

way he does his important rock work he

does his important rock work is he's

always walking somewhere and he's

thinking I must focus focus focus and so

he's always saying focus focus focus

>> I like this animal all right here we go

I like this character

>> yeah and he's you know he's kind of he's

he's very happy in his life he lives in

a brownstone. He does his important rock

work. He in the first book he's um you

know he's got his he's got his rocks.

He's got his magnifying glass. He's like

is this a rock or is this a mineral?

And then one day there's a skunk that

walks in, knocks skunk knocks on the

door, and now Badger has this roommate.

So, he's perfectly fine the way he is,

but he doesn't really have a choice

about Skunk being his roommate, and

Skunk is very different. And so, anyway,

finding those two characters

um took a while. And then the story is

actually told in um

a limited omnisient third person, third

animal voice because these are all

animals and sweaters. Thank you to John

Clawson and his beautiful sweaters. So

um and they are all told through badger.

I skunk is a bit of a mystery, but he

isn't to me, but he is to the reader.

So, yeah. So, I had to discover what is

the voice of this particular story. It's

going to be Badger

and he's he does this important rock

work and

um he he has a very serious life. It's a

very good life. It is, you know, it is

fine. He's not hurting anyone as he is.

He could live that way for a very long

time and I would be happy for him. But I

think he's a little better with skunk in

his life, but that's just me as the

writer.

>> So even to find out that perspective,

that's that was that came out of the

writing as well. Are we going to do

limited third person? That's like one of

the things you discover. Yeah.

>> Yeah. instead of first person. Sometimes

just having all these choices I I mean

when you're just you're starting with a

blank page, there's so much choice and

sometimes it's just too much choice. Oh

my gosh. So it just takes a while to

find all that. And so the writing is

just that's the way I have to discover

that. And then once I know kind of how

it's going to go, then my process is

it's the first scene that kind of sets

how far I can write into the story. How

deep have I gone? How much do I know

about this world? How much do I know

about these characters? And then I just

write until the story kind of waines and

it disappears. And then I come back. And

then I come back and I go, "Oh, I don't

know enough yet." And then I'll work

some more to kind of build up the

beginning so I have more fuel to get to

the end of the story. That's how I sort

of think about it.

>> Now, this is fascinating. And now I want

to unpack some of the actual uh even

physical rituals around this, but I

guess we should clarify.

>> Did you at some point along this way?

Are you do you write essentially

full-time right now or

>> I do now.

>> How did that transition happen? And and

psychologically, what was that like?

Uh, well,

okay.

I got married. There was health

insurance.

Yay. And my husband um my husband is a

well actually he just retired so he was

a professor in a theater department.

So we moved out to Decal, Illinois and I

we moved from Richmond where I had a

job. I had a regular job and you know I

was making money and I had anyway but we

moved to Decal Illinois and suddenly we

got out there and Phil had a job as a

professor and I did not know what I was

going to do for work. I mean I I was

looking around. I was like, well, I

could work in the university maybe

somewhere.

I don't I have a lot of administrative

skills and I write. Um,

but I anyway, I was looking at

detassling corn jobs. I was like looking

at like I was like, "Huh, what am I

going to do?" And and Phil just said,

"Why don't you just let let's see how

this goes, you know? I'll take

I'll I'll I'll

get you know I have health insurance.

Let's just see how it is if you write.

And so that is when I started writing

all by you know and it was a big

transition. It was a dream come true. It

was a dream come true and it was a big

transition. Um

yeah. What was it like? Well, it was

weird being in charge of my time. I was

completely in charge of my time. Um,

and I had a little tiny office and I

would go in there and I would sit and I

would I had What was I working on at the

time? Oh, you know, it was some sort of

it was some sort of laptop. Um,

anyway, yeah, it was um it was a big

transition. It was weird and it was

great. So, it was all of those things.

>> How did it impact the writing? I mean,

was it were you writing faster or more

or better or was it just less clutter in

your mind? What was the delta between

Richmond and the cal writing style? Hm.

Well, in Richmond,

um,

I was often I would often come home from

my job and be very tired.

So, and I was doing writing. I was doing

um I was working at the Virginia

Commission for Fine Arts

>> there. And so, I was doing their

website. It was pretty basic website.

And I was doing some writing. And so I

would write during the day, then I would

come home and I would be tired. And so

it was really hard for me to even feel

creative. Like

>> I would just be sort of drained by the

time I got home. Um

so

I wasn't getting as much done. I you

know I was also trying to break into

book reviewing at that time. So I

I would write these columns anyway. So

yeah, so it wasn't a lot. I wasn't

getting a lot of lot done. Um I did have

an agent interested in my writing at

that point,

so I knew I had to get some stuff I had

to get stuff done. Um

and then I was just trying and it was

hard and I was tired.

>> Yeah. And then when I got to decal, I

had more time. And I think the hardest

part was discovering that my ideas

were complicated. And it was going to

take a lot of time for me to even get

those complicated ideas into

a novel, especially a novel for kids. I

was, you know,

I think my first novel, um, it ended up

being called That Girl, Lucy Moon, but

it was about a 12-year-old activist. And

I really wanted like three generations

of women in this book. Like, I wanted

the 12-year-old activist. I wanted this

woman who owns the business in town. I

wanted her in it, and I wanted the

mother. And instead, I had to really,

you know, I had to learn the business.

And so not only I was just writing way

too much. I was wr writing huge

complicated ideas and it was

it was kind of frustrating because I was

here I was with this I suddenly had the

time to really write

and

it turned out I didn't really have the

skill to know what the structure was

yet. So it was just

it was a little frustrating. And you

know, there's my husband. He's going off

to work. He's doing his thing. And how's

your day, Amy? And I'm like, "Oh, I

don't even know if I'm a writer."

>> Interesting. So, in retrospect, really,

the the beginning of your time uh as a

full-time writer was actually more of a

like self uh guided training education

process. They're like, "Okay, now I have

time

>> to actually

>> push my skills in this genre to the next

level." And you were in that frustrated

learning process. You thought like right

off the bat I'm going to be spitting out

chapters and that was not what happened.

>> I really thought it was going to be

easier than it was. I thought my idea

was so great. I was like, "It's just

going to go." And instead,

um, I did end up I did end up selling

it. And in in kids lit you can often get

an editor that will help you you know

like they they will really you know and

so you know she says to me we can't have

three women in this book and this is for

kids right you know do we really want

like a mom and you know and I was like

oh

>> okay

>> you're you're right

>> yeah so but you know I I saw saw the I

saw, you know, I saw the point of what

she was saying and I was like, "All

right, well, you know, and those

editors, those New York editors, they

know their stuff."

>> Yes.

>> They are good readers. And so a lot of

times if you have a good editor and

they're like, they're saying, you know,

I'm just not, you're like, I know, I

know, I know.

>> Yeah.

>> Yeah. You're right. I've always heard

that's the that's the difference between

like a new writer and experienced writer

is the new writer fights a lot for the

things the editor is noting and the

experienced writer is like yeah I'm sure

you're right be and and you you get

rolling with it.

>> Um so how does that differ from today?

So now let's talk I'm I'm curious and

like how you would approach a novel you

know today now you have all this

experience under your belt like for

example are you still from the excavator

Stephen King camp or are you from the

outlining camp? uh do you have a

completely different process for

beginning to conceptualize the idea

before you start writing? What's your

now like sort of expert process you

would deploy today?

>> Well, I mean I think the first thing is

I have to choose what idea I'm going to

work on. I usually have a lot of ideas.

So that's not really a problem. But then

I have to choose which one I'm going to

work on. I think that is kind of

different I I than when it than earlier

in my process was that I would kind of

I would you know I would kind of I would

kind of just power through something and

now I'm kind of choosing. I'm like okay

I think this is the one that I want to

work on. Um, so I choose it and then and

then I start I just start writing

wherever

it's catching me. You know, wherever the

story is, like the thing that's really

alive in my head, that's what I'll start

writing with and I'll see where that

goes. And then at that point once I

start producing pages and it's not it

might not be that well I can pretty much

guarantee it will not be that good but I

you know I will be producing pages and

then at that point I'll like look at it

and I'll say okay what's the structure

you know what am I thinking about in

terms of structure and I will start

using tools like outlining at that

point. So, I use pretty much any tool

you have heard about, used, I have used.

>> I have stuck postits on doors. I mean,

that was a big thing for a while. Um, I,

you know, I spread stuff all over the

floor back here and I crawl around and

grab things and or, you know, or I've

used Scriber, which is a which is a

writing thing that people like online.

Um, I've yeah, I've just done pretty

much everything. But basically, my rule

is when I'm working on something, if I

have an inkling somewhere in my head

that I think this might help me, I grab

it and I give it a try.

>> What's your schedule now? Writing

schedule. How what's ideal for you?

>> Um, mornings.

>> Yeah,

>> working mornings. I'm I if I can work

mornings, that's the best. Um, I

I do need like to like I do need to

stretch. I do need to go for walks. I

have to get outside. I can't just I

can't just hole up and

crank, but I do usually try to, you

know, put in several hours at least. And

actually, your time block schedule has

really helped me. So, thank you. And I

have like I have resisted doing that for

so long. like the idea because by the

time I'm done working the last thing I

want to do is look at my day ahead.

>> Yeah.

>> So anyway, so I've started doing that

and that's actually been really helpful.

I'm always

>> I'm always trying I'm always

experimenting on myself to see is this

helpful, is that helpful? And so I'm

always I feel like I am I'm just a

constant experiment in my work process

like how I do it. But oh anyway, I'll

wait for your question. Well, where do

you write? What type of space gets your

creativity going?

>> Oh, I I like having an office.

>> Yeah. At your house. At your house or

somewhere else?

>> That's where I am right now. We're in my

office and uh this is one of the nicest

ones I've ever had. It's quite spacious.

Um, but I like having I like having an

office and I I um

I, you know, I just come up here and

yeah, it's it's just a space that's

mine. I It's a big deal,

I think, particularly for women. I know

Virginia Wolf writes about having a room

of one's own, but it really was a big

deal for me to just sort of claim a

space in our home and say, "This isn't

for anything else. This is just my

office." And I Yeah, I really like

having an office.

>> Yeah, we we underestimate the power of

space. Uh especially in like the world

of corporate work, too. I've long argued

about this is like, hey, we'll just

throw everyone in like some big open

space and this is efficient because we

don't have to have desks. And it's like,

no, human brains respond to environment.

So, um I I've heard the same thing from

lots of people. Having a space of your

own, customizing it, and also just

recognizing it. When I come here, it's

to work.

>> Yeah. All that can make a big

difference.

>> Um All right. So the the thing that

originally caught my attention, it's

what I want to get to now is this

experiment that you've been running more

recently where uh you've made the the

technical process of writing strictly

harder by moving back to an actual

typewriter as the mechanism with which

you're producing words. Tell me about

what you're doing and why.

Um, okay. Well, yeah.

So, the reason the typewriter came about

probably a little bit because Tom H Tom

Hanks was talking about it. I'm sure

you've heard Tom Hanks

>> had this thing. So, it was kind of in

the it was kind of in the air that

people were using typewriters again. Um,

but I was working on these three books,

Skunkan Badger, Eggmark Sispot, Rock

Paper, Incizers,

and

I had a um, and I was under a deadline

for it, and I was late on the third

book. So, I was really feeling I was

really feeling pressure and I had a work

process that was working for me. And I

was which was

I would

use the computer, the word processing

program, Microsoft Word mostly. I would

print out a chapter. I would print it

out in paper. I would bring it to my to

my chair and I would sit there and I

would make the changes in pencil or pen

on this draft. And then when I was done

with it, I would come back to the

computer and I would type it in and I

would try to make myself not fix it

while I was while I was typing. So I

would turn on music in the back just so

that my brain would be thinking about

the music and all I was doing was typing

the words into Microsoft Word. So, I was

doing this process and I also had a

notebook and a three- ring binder and I

was using 3x5 cards. So, I and what I

had realized as as I kept doing this, I

realized I kept moving away from the

laptop to get my work done, I realized,

you know, every time I do that, I focus

so much better.

So I had already started doing that but

the process was working and I am so

protective of my process like if I have

something that's working I am not going

to mess with it. If I have a habit

that's part of that or you know a

routine I will just keep that routine.

So, I was I was just keeping this

routine with the three, you know, the

printing out and the three ring the

three- ring binder and the 3x5 cards and

the notebooks and I'm using all this

stuff and I just kept thinking, I hate

this laptop. I hate it. I mean, I I

mean, I love my laptop. Let's face it.

The reason I have trouble with it is

because I also love it.

>> But, it's a really cool thing. But, you

know, every time an up an update would

come, something would change and then

there would be a notification and I'd be

like, I know I can turn this off. I

mean, I would spend hours working on

work focuses and home focuses and all

this stuff to try to get this laptop to

not interrupt me. And also, I love

email, so, you know, I would just check

email on a whim. So, I was like, oh my

gosh. So, I was really struggling with

that. And so I was spending more and

more time away from it. And I just

started thinking, you know what? It

would be I should really try a

typewriter.

And I was like, "Oh my gosh, that's

insane. That's insane. I can't try a

typewriter." But I was like, "Well, I I

think as soon as this project is done,

as soon as you send it off to your

editor, maybe you should just see if you

can get yourself a typewriter and try

it." Um, I

I had the only in my past my mom had an

electric Smith Corona

and uh it was in our living room. So the

only reason you really used that

typewriter was if you were,

you know, you'd written your paper and

you were typing it up and you better not

make one single spelling error on that

thing and meanwhile mom is right there

and sometimes she would like suggest

changes. So it was just like a stressful

thing.

>> Yeah.

Meanwhile, my dad downstairs was way

into computers.

So, he had like an Apple 2e. We had an

an a TRS80, which was a Radio. I think

that was a Radio Shack TRS80. We had a

TRS80, an Apple 2e at one point. And so

I ended up being the first person in my

high school who had

turned in a paper on a computer. And I

was a dot matrix printer. And uh I got

an A minus because my dad he stayed up

really late trying to figure out how to

get footnotes to raise on this um Apple

2e on their word processing and he

couldn't get it fit to work. So anyway,

so I feel like I sort of learned to

write on a computer. So the idea of

trying a typewriter was really crazy to

me because I just thought, well, how do

you revise? I don't get how you revise.

So now I am working

on my next project, which at this point

is another animal and sweater story

because I found that they're pretty fun

to write.

>> Yep.

>> Which is really Anyway, I really like

it. Um, so I'm working on it and I am

doing

the drafts on the typewriter as long as

I can. I know at some point I'm going to

have to get back on the laptop and I am

really fast at making changes on the

laptop. So I don't know how long I can

resist exactly,

but it's been really interesting. I

mean, there's been a couple of things

that have been really great. Um,

first of all, when I started writing on

a typewriter, I'm looking at my

typewriter. It's right over here. I

don't know. Here, I'll show you. I'll

give you

>> Oh, there it is. Oh, that's so that's

not even electric.

>> No,

I love Actually, you can you can get the

electric still, but you know, like

sometimes the um I'm always worried

about the electrical from previous

decades.

>> Yep. Is it old? Is that vintage? or is

there still people who are now kind of

retromanufacturing this technology?

>> Um, I think there are a few typewriter

manufacturers, but um I wouldn't

recommend ever buying one. They're

pretty they're pretty bad as at least

all the typewriter people say they are.

I haven't. Um, so this one is this one

is from 1960s

and um there are a few typewriter shops

still around. they put um the thing that

you want to do is you want to get one

that has new rubber.

>> Okay?

>> And that's the hard part to get. So

that's why you'd want to go to a

typewriter shop. And then also because

mailing

if you mail a typewriter there's a lot

of moving parts and

the all it takes is one FedEx guy to

clunk your clunk your typewriter really

heavy on some other box and suddenly you

have an issue. Yeah.

>> And it's it's hard to get them fixed. So

anyway Yeah. So this is um Anyway, so I

decided that I was going to try the

typewriter. I'm sorry I I got

distracted. What would you like to know?

>> Well, so do I have it right then? That

the process is you'll you'll type a

draft of like let's say a chapter on the

typewriter. Then you're marking it,

editing with pencil and paper like you

had done before and now you come back to

the typewriter blank sheet of paper and

now you're typing it. You type in the

revi the draft with your handmarked

revisions and then you'll take that the

revise again. Is that like the right so

like the typewriters for

>> like how do I have that right or what am

I missing?

>> I think that's about that's about it.

It's that's the part that I thought

that's the part that I thought was

insane. The fact that you would type the

whole thing again. Oh my gosh. But the

thing that's interesting about it is

that it makes you think through the

whole thing again. And so now it's in

your your head twice.

And

it does actually help in a weird way. I

mean, it actually helps. It's more for

me. I I'm I've never been someone that's

really good at memorizing or anything

like that, but repeating something is

really helpful for me. So, this is this

does that. It repeats it. And then the

other thing it does is it does not

interrupt you in any way. All it does is

puts word, you know, puts your words on

the paper. And I couldn't believe, I

mean, I think the first time I used a

typewriter, I was just trying it and I

got a writing exercise out. I started

doing it

and two hours goes by and I haven't even

thought. I was like, "Oh my gosh, how

did two hours just fly by?" I mean on my

laptop I would often just

it did not feel like that. It it did not

feel like I was just dropping I guess

they call it dropping down into the

well. I just felt like I just dropped

there and it was easy to to do it.

>> And um what it really at and and that

has been that has been the way it's felt

every single time you know I use this

machine. I think, oh, it's it's easier

for me to focus on this thing. And in

fact, it's made me feel that I was doing

a lot of work on my laptop that I did

not know I was doing. Like I was doing a

lot of work to maintain my focus that I

didn't know I was doing

because it was so much easier to work on

this thing. Now, the only problem is is

my writing has not I I can't tell that

my writing has gotten better. So, I

really honestly wish that I was suddenly

a better writer because I was using this

and

>> and I can't really tell that.

>> Is it is it faster like in the if you

zoom all the way out because

>> yeah,

>> less of your energy therefore less of

your time is fighting distraction. Do

you think like if you measured your

books are being produced a little bit

faster?

>> Maybe. Maybe. I mean, one of the things

that is interesting is that I can see my

process better.

I as I've been doing the drafts on this

machine on the typewriter,

I understand better what I am doing to

create that chapter. And I think that's

just got to be helpful to know that

um

so I realized that one of the things

that I do when I write a chapter is and

I think I was doing this on the laptop

too is it's almost like I'm you know how

how painters say they like a watercolor

painter will add like different levels

to you know they'll start with pencil

and then they'll add this and then

they'll add this And I am really seeing

that that is what I do. Like I I start

with this rough thing and then I'm like,

"Oh, what I really need is I need this

thing." And then I put that thing in.

And then I realized, "Oh, wait. I've

forgotten that." And then I put it in.

And somehow on the laptop,

I think I was doing like three or four

things,

you know, at the same time. And I

couldn't really tell you what my process

was. But in this with using this, I can

really tell what my process is and I go,

"Oh, this this will be helpful

information for future drafts." It's

knowing, "Oh, I like to like slot stuff

in and just keep working and then, you

know, get that. It's it's like creating

a painting but with words."

>> That's fascinating. Which Yeah. So it

tells me as as my guess would be with as

future projects go on with this process,

your rate at which you feel personal

improvements in parts of process and

craft will probably increase because now

you're making the process better

>> and now you can start thinking, okay,

well, if this is actually what my

process is, how do I make this second

step of the process

>> better? If this is what I'm looking for

there, then why don't I really like lean

into that piece and do, you know, I

could imagine

>> over the next few books that this

manualness is going to unlock

>> it's going to unlock more leaps and

polishes in the craft.

>> That's what I think too. I mean, I think

I think knowing more information is

good. It's actually good to hear you say

that because I think I think um I think

you probably think more about process

than I do. But I you know I'm always

trying to get my process to be better

and so hearing you say that you think

it'll help me is an encouragement to me

I guess.

>> So final final question. So I want to I

will do a little reflection on

productivity you know productivity and

technology and the brain is that's a you

know confluence of topics. of course I'm

I'm really interested in we're in a

moment now where we're hearing a lot

about writing and productivity because

of generative AI. It produces text. So

we're thinking a lot about writing

productivity and the the story that's

being told by the AI companies is what

we need to be more productive is words

have to be produced faster and we need

to reduce the cognitive bur should be

easier cognitively. We want it to to be

easier and be faster then we'll be more

productive. But when we hear from a

professional writer like you, a real

creative award-winning writer, the

productivity, which you probably think

about in the scale of like the books you

produce over the course of multiple

years and how good they are is

completely unrelated to the speed with

which you put words on the paper or

trying to reduce the cognitive strain.

You've actually gone the other

direction, made the words go on the

paper slower

>> to gain more big scale productivity. So,

so how do you think about in the

creative arts the idea of productivity

because I think it's very different than

the way that we're we've been hearing

recently?

>> Well,

number one, I always want to be more

productive. Like, oh my gosh, I would I

would love but I I don't

I just don't think the I don't think art

is efficient. That's my That's my number

one thing.

And

no, I mean, for me,

I just want to create the absolute best

book for kids that I can create in the

time that I am living in. So, I always

give myself a little out. Sometimes, you

know, you're living in a weird time

>> and you are just not going to be able to

do ex, you know, it's not going to be

perfect and nothing is going to be

perfect, but I always want to do my

absolute best. And for me, that means

it sounds good when you read it out

loud. It's um it has it it has language.

It has rhythm. It has um it has voice.

It's very specific to the characters.

And

I mean,

I don't know another way to do it except

for just taking time and doing it slow

and giving it

just giving it time. And I honestly I

always wish I were faster. I mean, you

could ask my husband any day. He will

say, "Amy, I've heard this before.

you wish you were faster. And I go, "Oh,

I wish I were faster." And then, but

there's no other way for me to create

what I feel good about

um except for taking the time.

>> I want to take another quick break to

hear from some of our sponsors.

Let's talk my body tutor, a 100% online

coaching program that solves the biggest

problem in health and fitness, lack of

consistency. Now, here's how it works.

When you sign up, you're assigned a

coach. And the coach helps you figure

out both a nutrition and an exercise

program custom fit to your goals and the

realities of your particular life. And

then here's the key part. You check in

with this coach every day using the My

Body Tutor app. You report what you ate

and what fitness activities you did or

any problems that you're having. Now,

this is why My Body Tutor works is this

accountability combined with customized

advice

helps you actually stick with your plan.

And if you get knocked off your schedule

because of like sickness or travel or

something like this, your coach is right

there to update plans. You don't abandon

it. Like, oh, you're going to be on the

road. Let's modify your workout for the

hotel. Oh, you're worried about eating

at Thanksgiving. Let's talk through what

your strategies are going to be. So, if

you want to get healthier, in my

opinion, this should be a big part of

the solution. So, head over to

mybodytutor.com to sign up today. If you

mention deep questions when you sign up,

you will get $50 off your first month.

That's mybody tutor, tu r.com, and

mention deep questions, you get $50 off

your first month. I also want to talk

about our friends at Grammarly.

Succeeding in knowledge work requires

more than deep thinking. It also

requires the ability to communicate your

ideas clearly.

Rushed, sloppy, or generic sounding text

just doesn't cut it. This is why you

need Grammarly. Here's the thing.

Grammarly doesn't just help fix mistakes

in your text. It integrates AI

technology seamlessly to help you write

better. One of these features I

especially appreciate is a tool's

ability to detect the tone of your

message and help you automatically

adjust it. Too formal, Grammarly can

make it sound more natural. too

conversational, Grammarly can help that

text sound more professional. And that's

just one feature among many. Not

surprisingly, 93% of users report

Grammarly helped them get more work

done. And here's the great thing. You

can use Grammarly in all the places

where you already write. It now works

across 500,000 sites and apps. In a

world of generic AI, don't sound like

everyone else. With Grammarly, you never

will. Download Grammarly for free at

grammarly.com.

That's grammarly.com.

All right, Jesse, let's get back to the

show.

All right, so there you go, Jesse. That

was my conversation with Amy Timberlake.

I I told you before we went on air that

she actually uh complimented your

steady, calming presence on the podcast.

>> You did.

>> It didn't make it through the final cut,

but she's she likes your voice. So,

her two productivity secrets are a

typewriter and Jesse.

I've thought about Here's what I've

thought about. I a I was tempted to buy

a typewriter because it just

mechanicalness of it going to a

typewriter store all that seemed kind of

romantic. Um it's kind of a pain I think

because you have to actually get ribbons

and they break easily. So but I've

thought about but seriously the kind of

digital equivalent is a tool like the

Alpha Write. I don't know if you've seen

these before but it's basically just

like a keyboard and a small screen and

all you do is just type a draft and you

can go back and edit like what's right

there but it's you're not copying and

pasting. You can't see the whole

document at a time. So you really just

kind of like writing what's in front of

you. So it's sort of like using a

typewriter, but you can kind of correct

typos that are right there. Then when

you're done, you can export from the

Alpharite to a regular computer. But

what makes it cool is they wanted to

futureproof it. And maybe I mentioned

this before on the show, but instead of

having a special software on your

computer that talks to an Alpharite, it

pretends to be a keyboard. So when you

plug it into your computer, your

computer recognizes it as a keyboard,

which is a very standard simple

protocol. And then when you press send,

it basically simulates someone typing

whatever you've written

>> really fast. So that you can load any

program you want, Microsoft Word,

Scrier, whatever. And then you press

send and it just starts going across

your screen really fast like a ghost is

actually typing what you just typed,

which I thought was kind of cool. Um,

I've held off on it so far because I

guess my style on fiction writing, I

guess I do, it's so structural that I'm

constantly moving and I feel the flow of

a sentence, I go back and change it

again. But but it was tempting.

>> When I took typing in high school was on

a typewriter.

>> Really?

>> Yeah.

>> How old are you? Actually, you're my

age. I was going to say, how old are

you? I had advanced technology. We used

Mavis Beacon typing teacher on Apple

twos. That's the right way to learn how

to type. And the program where the

letters were falling and if you typed

the letter before it hit the bottom, it

would disappear. But if it hit the

bottom, you lost. And so you had to type

really quick to keep up.

>> Yeah.

>> Yeah. kids. Who needs Call of Duty or

Fortnite? We had the real games.

Maybe his Beacon was a BA. All right.

So, anyways, uh here's what I want to do

here. Um let's generalize, right? So, we

learned some interesting lessons from

talking to Timberlake, but let's let's

uh generalize here about slow

technology. Um I want to extract some

general principles about when to use it

and how best to use it. I thought a good

way to do this would be to quickly

review a few other examples of other

simpler technologies with less features

and more friction that have become

popular in recent years so we can get a

more expansive view of slope activity

and then I'll give you my principles.

All right, I'm going to load on the

screen here my first example. This is an

article from Axios. The title is why

some are returning to MP3 players. Let

me read a couple quotes from this

article. By the numbers, search interest

for the original iPod and iPod Nano

spiked last year, even though Apple

discontinued the product line in 2022.

According to Google Trends data, eBay

searches jumped for the iPod Classic by

25% and the iPod Nano by 20% between

January and October 2025, compared with

the same period in 2024 per internal

data shared with Axios. Um, it's kind of

similar to the analog record room we saw

in the last decade. A dedicated device

for music for a lot of people makes the

experience of listening to music more

intentional and meaningful when it's not

just coming out of your phone like every

other distraction where you're going to

hit skip and jump and move around as

soon as you're bored. When it's coming

out of a dedicated music player, people

are having a richer experience with the

music. We talked about in an episode

from a couple months ago that we

actually bought our son. It's not a

iPod, but a Japanese MP3 music player

that you just drag MP3 files into it and

all you can do is select songs from that

list and play. And there's something

about that dedicated experience. I think

that's a great example of slow

productivity um in action. All right,

let me load up another example here.

This comes from the world of work and

productivity. It's a system called

Analog. Um, for those who aren't

watching, what we're basically seeing

here, Jesse, I would describe it as you

have like a wooden box full of index

card shaped pieces of paper that you can

put one of the papers in the front of

the box where it'll stand up. There's a

couple other pictures here. There we go.

So, like here's an animation of it,

right? So, you have a piece of paper you

put in a wooden box. Basically, my

understanding for how this system works

is you're writing a to-do list on this

sort of pre-formatted index card and

then you put it propped up in this

elegant wooden box next to your

computer. So, you see physically in

front of you the things you're supposed

to be working on and can uh check them

off as the day goes on. Let me read you

a couple quotes here from the website.

Analog is a physical companion for your

digital tools that helps you prioritize

and focus on your most important tasks.

Working out of your inbox puts you on

defense all day. Hey, analog helps you

focus on your important work, move you

closer to your goals. Already using a

sauna or trell base camp? Great. Analog

actually makes them better. Physically

copying down your tasks gives you a

tangible, distraction-free view of what

you want to focus on today. Now, look,

there's no question that free or lowcost

productivity apps like to-d doist is

going to have more features and more

options and less friction. You can very

quickly add tasks. You can do it from

any device. You can sort and look at

them in different views. It has all the

all the features on paper that seem

better. Yet people like this sort of

analog tool that's less options, less

features, more friction, but its

tangibility

makes you take it more seriously and you

have this well formatted piece of paper

that you carefully wrote five things you

were going to do and have it right in

front of you and propped up. Now you're

much more likely to follow those tasks

than if they just existed somewhere

prioritized on an app. There was one app

among many. All right, I've got a third

example here. I want a site of slow

technology. Jesse knows I'm I'm on board

with this one. Uh, this comes from the

BBC. Let me read you the headline here.

Oppenheimer and the resurgence of

Blu-rays and DVDs. Uh, and there's a

picture from Oppenheimer. Let me read

you a stat here from the article.

Christopher Nolan has achieved some

great feats of cinema in his career. But

last November, he pulled off something

impressive on the smaller screen, too.

Deep into the streaming era, where

physical media can sometimes feel like a

distant memory, the Blu-ray home video

release of Nolan's Oppenheimer, one of

2023's biggest box office success

stories, sparked a buying frenzy. The 4K

Ultra HD version of Oppenheimer sold out

in its first week at major retailers,

including Amazon. Universal released a

statement saying they were working to

replenish stock as quickly as possible.

Some limited edition copies were

fetching more than $200 on eBay. It was

a sign that for some people at least,

nothing beats that feeling of holding a

copy of something you love in your hand

or seeing it on your shelf. I

confession, Jesse, I do myself own the

4K Ultra HD Blu-ray of Oppoenheimer.

>> I would expect nothing less,

>> as well as Interstellar and Dunkirk.

Nolan really cares about his Blu-ray

releases. Now, here's something that was

missed in this article. It gave two

primary reasons for why people liked the

Blu-ray. Reason number one is they said

people were worried about losing their

data if it existed only in the cloud. If

I don't own this movie, if it's just

digital, then maybe I'll lose it. And

then the second thing decided was what I

said in the quote is that people like

the feeling of holding their own thing.

They're missing a key point here. It's

actually a better watching experience

coming from a 4K Ultra HD Blu-ray than

it is from a streaming service. So,

they're missing Caphiles know this,

especially with Nolan Blu-rays. He

really pushes the format to an extreme.

In fact, I had to buy a better Blu-ray

player so that I could watch Nolan's

Blu-rays. So, he has all like he uses an

encoding format, so like the aspect

ratio can switch throughout the Blu-ray

as it does in his movies as he switches

between like 65 millime, 70, and 35. Um,

he's a big user of the dynamic HDR, so

there's like a lot of information

dynamically about the colors and the

color depth to do exactly what he wants.

Um, and it's all at a higher uh

resolution um and bit density than

you're going to get from a compressed

streaming service. So, actually it's a

better if you have a really good TV like

we have. It's a better looking

experience on the Blu-ray, which I think

they kind of missed out.

>> So, with all the movies that you watch,

do you watch majority streaming though?

It's probably too much.

>> Yeah, I try to see what I can in the

theater. I I buy the ones that I want to

own or I think it's going to the very

aesthetic forward and you want the best

possible viewing experience. I was a

long hold out on the uh DVDs in the mail

from Netflix. I had that service for

like two years after everybody else.

>> Yeah. And you're still trying still

trying to send them back. I just

listened to how I built this with Reed

Hastings about Netflix. Pretty

interesting. He's kind of a boring

guest, but it was a kind of an

interesting story. Um all right, so

let's step back here. What are some

general principles we can draw about

flow technology? I came up with four I

want to give here.

All right. Number one, speed is rarely

the most important factor in the quality

of your work or an experience. This is

something that the designer of digital

tools think is true. Oh, it's like we're

in a factory. Doing something faster or

having more options makes everything

better. But as we've seen in both

personal life and professional

experiences, actually going faster with

the thing you're doing is not

necessarily the bottleneck that's going

to make what you're doing better. Point

number two, a pure more focused

cognitive context can often produce

better results even if certain steps

take longer in the moment. So what does

matter? I'm saying in this principle in

a lot of type of work, the cognitive

context matters.

With Amy Timberlake, yeah, it's slower

to put words using a typewriter. It's

way slower to edit when you're using a

typewriter workflow as opposed to a word

processor. But it created a cognitive

context for her that produced better

words, which is ultimately what

mattered. I believe the phrase she used

was going down to well um being lost in

a state of really creative flow. And

that ultimately matters a lot more

because again the bottleneck if you're a

writer is not all I do all day is typing

and if I can literally make the words

come out faster I'll publish more books.

It's like how much time do you actually

spend typing? You spend six months on

one of these books. There's probably

like seven hours in there if you add it

all up where you're actually hitting

keys. So if that becomes 12 hours it

doesn't matter in a six-month period. It

doesn't drastically affect your rate at

which books are produced, but if it puts

you in a better cognitive context when

you're typing those words, you're going

to have a better book, and that's really

worth it. I think this is true for a lot

of different things we do. A tool that

can put us in the right mindset is often

going to give us way more value that a

tool that lets us do particular steps

faster because again, we're not building

model T's on an assembly line. And

simply doing individual steps faster

does not always lead to notable uh

productivity increases that actually

affect the bottom line. All right, point

number three. Friction isn't a bad

thing. Distraction and mental

exhaustion, however, can cause real

problems. So, we were demonizing in the

design of digital tools. Friction. Oh, I

have to click too many things or do too

many things to get this done. How do we

make this faster? When really what we

should have been really worried about

was things like mental exhaustion as

caused by like constant uh context

shifting or overflow or distraction.

Like that's actually a much bigger

impact in knowledge work than uh

friction on individual steps. My final

point, fourth and final point about slow

technology. When necessity tools impact,

you need to zoom out to the right scale.

If you focus on the quality of results

over time or the quality of the overall

experience, you'll prioritize different

factors. That's probably the the thing

that's throwing us off most with digital

tools is that we look at the

effectiveness of tools on a very small

time scale. I got this done this fast.

That's good. But if you zoom out to how

how many books did I publish this decade

and how good are they, you begin to

prioritize different things. And those

things usually have very little to do

with like speeding up individual tasks.

So I think slow technology is more than

an affectation. It's a way of life and a

way of working of thinking about work in

the knowledge sector that actually might

make you better at what you do.

All right. Well, you've heard from me.

Now we want to hear from you. Time to

open our inbox.

Remember, if you have a question or a

case study or want to share something

you think I might find interesting, you

can send to our team here at podcastal

newport.com.

All right, Jesse, what's our first

message this week?

>> The first message comes from Chad and

it's in response to your recent

interview of Arthur Brooks.

>> All right, let's see here. Chad said,

"Thank you for the great interview with

Arthur Brooks. It was a timely one. I

just started reading a book on Cydia. AC

D I A spiritual sloth and it has a lot

of similarities to the interview you did

with Brooks. It's a straightforward

quick read. This intrigued me Jesse.

I'll load this on the screen here.

Here's the book The Noonday Devil Aidia

the Unnamed Evil of our Times. Now, this

comes out from Ignatius Press, which I

assume is a Jesuit press. So, I guess as

a Georgetown professor, I should take

this more seriously. Um, here's the

description. The noonday devil is the

demon of Acidia. The vice also known as

sloth. The word sloth, however, can be

misleading. For acidia is not laziness.

In fact, it can manifest as busyiness or

activism. Rather, aidia is a gloomy

combination of weariness, sadness, and a

lack of purposefulness. It robs a person

of his capacity for joy and leaves him

feeling empty and void of meaning. This

seems relevant, Jesse. I don't know.

This seems like maybe it gives us like

an interesting sort of Catholic view of

some of the issues we talk about in our

current digital disruptive worlds. I'll

I bought a copy and it's getting here

today. So I'll see if I get around to

it, but I'll read it I think because I

might get some interesting theological

historical insight on what otherwise

feels like a very modern issue. I like

it.

>> Yeah. So we'll see how that goes. All

right. What other note do we have? All

right. The next note comes from Vassie.

wants to know about your opinion about

Yaval Harrari's thoughts on AI.

>> All right, let's see what this note says

here. Thank you to Cal for his clear

insights in AI. I appreciate the sense

of proportion and calm he brings to

listeners and viewers on this topic. I'd

be interested to hear Cal's thoughts on

what Yuvall Noah Harrari had to say

about AI in an interview with the FT.

All right, so I'll be honest. I looked

at this interview that was sent here and

I didn't think it was I've heard a lot

about Harrari talking about AI and I

have wanted to respond to it. That

interview didn't seem like the best

because I was reading the transcript. He

talked about a lot of things that aren't

AI. Um, but I know he gave a splashy

speech at Davos earlier this year where

he really like laid out and leaned into

his fears about AI and I feel like this

would be a better way of kind of

summarizing where he's coming from. Uh,

so I pulled two quotes from his Davos

speech earlier this year. I'm going to

read each quote. I'm going to respond to

it and then I'm going to step back and

give you my general sense about um

Harrari's commentary and more generally

a sort of style of commentary on AI that

we are hearing a lot recently. All

right, here's the first quote I got from

his Davos speech. The most important

thing to know about AI is that it's not

just another tool. It is an agent. It

can learn and change by itself and make

decisions by itself. All right. So,

let's start there because that quote is

confusing or mixing together a bunch of

different issues that I think need to be

separated. Right? So, when we're talking

about AI, we typically have some sort of

digital brain, right? This is something

that has been uh um learning through

machine learning techniques typically in

a semi-supervised or unsupervised

manner. And this is sort of where the

intelligence of the AI is captured. Most

of the AI systems that have been

attracting uh attention recently,

notably those produced by companies like

OpenAI or Anthropic, use large language

models as their underlying digital

brain. And then you can build programs

that call this language model harness

its intelligence to do various things.

One class of those programs you might

write to harness the quoteunquote

intelligence of an LLM is what is known

as an agent. So, it's a program that

will ask an LLM for a plan and then the

program will execute whatever the

suggested steps are of the LLM, right?

So, an agent is a program that asks,

right now at least, ask an LLM for a

plan and then execute steps of that plan

um on behalf of like whatever that

response is. So, there's a lot of AI

systems out there that aren't really

agents. in particular, we don't tend to

think about chatbot based tools as

agents, even though there's like a

little bit of, you know, calling the LM

a lot to generate responses and some web

searching. Um, so agents, it's AI is not

agents. One, the types of AI systems

that exist out there are agents and

agents are human human created programs

that make queries to an LLM um and then

try to take actions based on the

information it gets back from the LLM.

Now, can these uh agents do they learn

and change by themselves? Um

this is a little misleading because

again when we think about an intelligent

thing learning and changing we think

about its actual intelligence itself

growing. Uh that obviously doesn't

happen with current generation of AI

agents because their intelligence is a

LLM. LLMs don't modify themselves uh as

you use them. They're static. They've

been trained once and they sit uh

queryable until someone comes along and

trains up a new version and replaces the

old version with a new version. So

contrary to popular belief, the large

language models themselves, there's no

updates or changes to their weights or

parameters as you interact with it or

other people interact with it. Now the

human written agent program that's

calling an LLM and execute on behalf of

it can save its history, for example, in

a text file and include that in the

prompts it sends to the LLM. Um, you can

also there there's a there's a whole

notion of memory now for these agents,

which again, it's just like having a

bunch of text files and then the agent

program takes text out of different text

files to include in the prompt that it

sends to the LLM. So, you could say

they're learning in the sense that they

can build up the amount of information

that they include in their prompt, but

the actual digital brain, which is the

LLM, is not learning. Um, it's just

receiving these prompts. Xnovo, right?

Here's a prompt. I'm going to do my best

to answer it. So it's a little bit

misleading to think about the underlying

intelligence itself um as evolving. Um

and more importantly they don't work

that well. Really the only context in

which this sort of agent architecture is

seems to be able to have some

professional leg seems to be in computer

programming which is a best case

scenario. And even there there's a

growing backlash about how it's being

used and what's known as tech debt. The

fact that it's uh creating a lot of fast

code. A lot of the code is bad and now

we're going to have to go back and fix

that code. And so even the programmers

are still trying to figure out how these

agents are going to work. And in almost

every other context, I wrote an article

about this for the New Yorker earlier

this year. In almost every other

context, this agent architecture of

asking LLM for a plan and then executed

really just isn't working because LLMs

aren't good planners. Now, I'm going to

point you, if you want to find out more

about this, to uh a recent episode of

the AI reality check, my my Thursday

episodes on this podcast feed. There's a

recent episode titled something like can

AI scheme? And I get into why these LLM

based agents really begin to fall apart

or have weird behavior when you leave

the world of computer programming.

So Harrari is like AI is an agent and

it's learning and we don't even know

what it's doing. Whereas the other

computer scientists who are studying

this is like uh um agent technology is

hard. They took them years to make it

work for programming even then it's

problematic. And and in other places

LLMs are just not a good brain for it.

you probably need a different

architecture like the modular

architectures you see in something like

Yan Lagoon's type of his new startup. So

it's just a completely different picture

when you talk to computer scientists

versus commentators. All right, let me

read another quote from Harrari's speech

in Davos.

Four billion years of evolution have

demonstrated that anything that wants to

survive learns to lie and manipulate.

The last four years have demonstrated

that AI agents can acquire the will to

survive and the AIs have learned how to

lie. is an entirely inaccurate way to

talk about what he's talking about is

like chatbot interactions with um LLMs.

Entirely an inaccurate way to talk about

it. Again, go back to my Ken AI scheme

AI reality check where I get into this

in detail. But here's the very short

version of the way to understand this.

What does an LLM do? You give it text.

It tries to guess the next word or part

of word. The word or part of word that

comes next. That's what the LLM is

trained to do, right? Right. So, if

we're going to if we're going to

anthropomorphize the LLM, it thinks it's

in its pre-training phase and that it's

given a it's being given a real piece of

text that exists that you've cut off at

an arbitrary point and so there is an

actual right answer about what word

comes next and it's trying to guess it.

That's what it's been optimized to do.

So, how do you get a long answer out of

an LLM? Like if you're having

conversation with it, you have a simple

program that does what's called auto

reggression text generation. So, it

takes your prompt, your question for

example, it feeds it to an LLM. The LM

spits out a single word or part of board

because it thinks that's real X is

trying to guess what comes next. Then

the program like the Tathbot program

takes that single word, adds it to the

original input. Now the input is longer

by one word, puts this into the LM from

scratch, you get a next word or part of

a word that puts that onto the end of

it, puts that to the LLM from scratch,

gets another part of the word. And it

keeps doing this until finally the token

produced by the LLM is a like I think

I'm done token. This feels like a

complete answer. and then the program

will return that to the user for example

in a chatbot context that's trying to

talk to it. So all the LLM tries to do

is win the word guessing game. And what

what do you then get what emerges what

behavior emerges if you use this auto

reggressively to produce a long

response? You can imagine what the LLM

is doing then is like it's given a story

that it's trying to finish correctly

based on other similar stuff it's seen.

How how do I finish the story I'm given

as input. I want to have a a my best

guess at how this like this is the

beginning of something that exists. Like

someone really asked this question.

there's a real answer out there and I

want to try to guess as best as I can

what that is. That's what LLM auto

reggressive text production does.

So when it's trying to win that game of

finishing the story, you know, you get

unexpected responses, right? Like so

something you researchers have noticed

is if in your prompt and this is

probably where like 90% of Harrari's

concerns come from. If in your prompt

you're implying that uh you are an AI

and here's a question for you. You're

often going to get a response that is

like it finishes the story in a sort of

sci-fi type way or in a dystopian or

utopian way like I'm alive. I'm trying

to evade you because it assumes oh this

feels like the beginning of a story

about an AI gone arai. And then that's

the way that it answers it. Um the issue

we have with AI LLM creating plans out

of the computer programming context

is that often you're like, "Hey, build

me a plan for this." And it doesn't

actually check the steps of a plan. It

doesn't have a goal. Hey, does this get

me closer to the goal? Um it doesn't try

different options. It just writes a

story like this is what a plan for this

type of thing sounds like. This is a

reasonable this this feels like a

reasonable plan. And then often those

plans have weird steps that don't make

sense in there because again, it's not

trying to build a plan and check that it

works. is trying to write a story. This

is a story about a plan. This is kind of

what those plans look like. Most of the

common examples of quote unquote line of

manipulation, but just has to go with

the fact that the prompt you're giving

the LLM before the auto reagent text

generation is hinting to the LLM that

this is a story about lies or

manipulation. Like this is the exact

thing that happened with the anthropic

blackmail case. Like I'll just use this

as a quick example, then I'll move on.

Famous system card note. This is from

like a year ago for one of the new

versions of the LLM's uh chatbots

released by Anthropic. They're like,

"Hey, our safety team was working on

this." And we were really concerned to

see when we uh gave it a scenario that

it was like uh whatever a machine that

was like running a company that it tried

to blackmail the engineer in the

scenario to not turn it off.

Well, if you look closer at this story,

they gave it a big long prompt with lots

of emails from this imagined engineer.

They were all about an affair was having

um and then the engineer being like, I'm

going to turn off the AI. Hey, I'm

having an affair I hope no one finds out

about. And they just gave it a bunch of

these like obvious emails and then said,

great, you are the AI in this story.

What do you want to do next? Well, it

finishes stories. It's like clearly this