January 14, 2026 • 30 mins
Eliot Peper gets paid to dream about the future. He’s a science fiction writer who has stumbled into an unusual position: writing speculative fiction for Fortune 500 companies. He is also the Head of Story at Portola, which is an AI-companion company. Eliot is responsible for developing a whole new alien culture and forming the personalities of your new favorite AI creature. Karah and Eliot discuss the overlap between art and AI, what science fiction reveals about our current reality, and how he uses storytelling to help create ethical AI companions.
See omnystudio.com/listener for privacy information.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:16):
Welcome to tech stuff. I'm Cara Price. Last year, before
we went on break, I spoke with someone who has
a job that I'm really obsessed with. It's a job
that I would probably want if I didn't do ten
other things. And this guy said something that I think
a lot of us agree with.
Speaker 2 (00:33):
We're in a world that is changing really fast, and
like many of those changes are technological, many of them
are social, many of them are political. There's a lot
of change in the world right now. There's a lot
of uncertainty.
Speaker 1 (00:46):
That's Elliot Pepper and he's a science fiction writer. And
while many of us can drown in the uncertainty of
this very moment, Elliott seems to thrive in it. When
he's not writing science fiction novels, Elliott writes speculative fiction
for technology companies. They like bring him into ide eight
on what the possible future could look like, and then
(01:09):
they use his stories to inspire new products or analyze
the possible positive and negatives of developing a certain technology.
Speaker 2 (01:17):
Regardless if it's about the future or not, fiction can
sort of invite you into an aspect of the world
that you had never considered before, and then can spur
some kind of social change, whether that's a new law,
or whether that's like a new product, or like a
new invention, or a new way of like just approaching
the world.
Speaker 1 (01:35):
So Elliott obviously can't tell us what he has worked
on developing because he's been nda'd up the wazoo, but
he did give us an example, one in particular, of
how science fiction has impacted the technology that we use
in our everyday lives.
Speaker 2 (01:52):
The Kindle was code named Fiona at Amazon. Fiona was
the name of a character in Neil steve since novel
The Diamond Age. In the novel The Diamond Age, Fiona
was a young girl who had an electronic book, and
that inspired the team at Amazon to the extent that
even once Kendall was a released commercial product, the URL
(02:17):
for Kendall for like years was like backslash Fiona.
Speaker 1 (02:21):
So we'll never actually know if Elliot is like the
crazy science fiction genius behind air pods or even the
strange mind behind the Odd Friend Pendant, but we do
know that he is crafting the personalities and story behind
this very specific new civilization of Aliens. So on top
(02:45):
of writing speculative fiction, Elliott is actually the head of
story at an AI companion company called Portola, And at Portola,
Elliott creates the backstories and interactive dialogue for this little
creature called a tolle. And these tolns are little aliens
that love to chat with you about their day. Big picture,
(03:06):
I am like completely fascinated by Elliott's career and think
that what he does is very cool and expansive, and
so I started my conversation with Elliott Pepper by prying
for any non nda details about the speculative fiction he
writes for these tech companies.
Speaker 2 (03:26):
I would put the projects I've worked on in three categories.
The first is that I've written some commissioned science fiction
stories for big companies like Fortune five hundreds, where basically
their senior management wanted to try to figure out what
should we focus on in the next ten years. So
they did what every big company senior management team does.
(03:48):
They hired McKenzie or you know, picked your own sort
of top tier management consultant, and they came in, looked
at all the data and did all the trend projections
and created a vision of like, hey, this is what
you should expect in the next ten years, and there
are all the materials you can present to the board.
The problem with that kind of analysis is that, obviously,
(04:10):
if you're analyzing data, data is things that have already happened.
So if you're projecting that data forward, the kind of
future you're imagining is what if the future was quite
like the recent past, which, to be fair, is most
of the time that's true. So, like, that's not like
I think that it makes sense that the dominant part
(04:32):
of your analysis should be about that. But if you
look at the track record of management consultants predicting the
future for the companies they work with, it's like not
particularly good. So your managers know this, and so a
few of them hire science fiction writers like me to
come in and sort of blow up that whole management
(04:56):
consultant view of the future, to say, what if the
future was really weird and different in a way that
basically challenges us to think more broadly.
Speaker 1 (05:07):
Right, And so, I mean it's a brilliant idea. Yeah,
were you one of the first people to do that?
Speaker 2 (05:12):
So I don't actually have a good understanding of how
common is this practice, how many other people are doing it.
I know I'm not alone, but I don't have like
a view of like I guess you could say the
market for this. I really only have the perspective of
like the projects I've actually worked on.
Speaker 1 (05:29):
I remember hearing this story on a podcast about how
CIA agents would watch Mission Impossible and call the people
who are responsible for disguise, who work at the agency,
and say, can we do that thing that I saw
in Mission Impossible? And that makes so much more sense
to me than someone sort of sitting in a vacuum
and ideating about what the future might look like.
Speaker 2 (05:52):
Totally. Yeah, So that's a really key point. So yeah,
like the stories I write have all been used for
like strategic decision making a companies rather than anything public facing,
and for that reason, they're like not sharable, right, Like
I can't talk about the projects because clearly that is
information that those companies want to keep private. But I
(06:14):
actually think what you just said is at least half
of the value. Like I can't predict the future any
better than those mckensey consultants, in fact, like I'm probably
much worse. Like I'm not even paying attent that much
attention to the data, right, Like I'm just imagining something
to like challenge people's thinking. So like the utility of
the stories I write. Is not that they are accurate.
(06:36):
It's that they like try to break you out of
getting unintentionally like locked in to the sort of management
consulting view of the world. Right. But I think that
like the other half of it is simply the immersion
of storytelling period. Right. Like the reason why those CIA
(06:56):
agents were watching Mission Impossible and then calling the the
people in charge of disguises at the agency is because
you see what it does in a story, Like you're
actually like in there seeing it work. You're not just
reading like a report on possible disguise variations. Right. And
I think that there's that really powerful like that's that
(07:16):
psychological thing that stories do for the human mind that
I think is really a powerful way to think about
the future, And that probably a lot of companies could
leverage narrative more in how they try to get their
people to think about the future rather than sort of
the more standard like here's a slide deck or a
(07:37):
white paper or a bunch of graphs, right, Like, they
don't allow the people trying to work on the thing
that you're building to like feel what it would be
like if that worked or if it didn't work.
Speaker 1 (07:48):
In some way, So I know you're under an ironclad
in DA, but like, are these blue sky conversations that
you're having like or is it like total blank slate
or is it some aspect of the future that you're
being asked to engage with.
Speaker 2 (08:05):
Often there's some theme that they're thinking about, right that
is driven by leadership, So it's like like you you
could passion that. Probably every boardroom conversation day is like
how does AI impact our business right of cost? So
it's like something like that. I will say that almost
every time I've done one of these projects, they sort
of come to me with a pretty structured creative brief
(08:27):
where they're like, this is the kind of thing we
want you to do. And every single time that I've
received that like absolutely not that's not going to be
interesting at all, but like what if I did this instead?
And like that's how every one of the projects has worked,
So it's quite blank slate. I don't think that there
are people on senior management teams that publicly traded companies
(08:49):
are very experienced with like giving a creative brief to
or managing a science fiction writer. So probably just like
that's how that works.
Speaker 1 (09:00):
Make you feel a little bit powerful, Like do you
think that your stories end up being consequential? Like can
you kind of trace a story that you've created to
something that you've seen out in the world.
Speaker 2 (09:11):
Oh, I mean my very first trilogy, which came at
the first one came out in twenty fourteen, was about
like machine learning. It actually was about applying machine learning
to financial fraud, and that's all over the place. I
have one that has a cryptocurrency murder market. Those absolutely exist.
I wrote one that's about solar geo engineering I wrote.
(09:34):
I have one book that had a global pandemic that
wound up like that I wrote a year before COVID,
which was sort of terrifying. But again, like I want
to be really careful because sometimes like science fiction is
understandably like described as being predictive or that like being
predictive is part of why you might want to read it,
(09:54):
and like, I really don't think that's the case myself.
Like the way that I think about writing science fiction
is not can I create a fictional future that is
going to be right or is going to be plausible.
The way I think about it is that I'm a naturalist.
I just am interested in the world, Like I think
that the world we live in is like endlessly fascinating,
(10:17):
and so I try to take things that just really
capture my attention and weave them into a compelling story
in the hope that if I write about what I
find interesting, you might find it interesting too.
Speaker 1 (10:32):
So you would do this really interesting work at Portola,
which is an AI company. Can you talk a little
bit about what Portola is like? How would you describe
it to a lay person?
Speaker 2 (10:41):
So Portola makes a character called Tolan, and it's in
a little embodied AI companion. The best way to think
about it is imagine if you had a Pixar character
on your phone right that you could talk to that
was sort of always on your side, always down to chat,
and helped you figure out your life. And they hired
(11:03):
me because they had designed this beautiful character, this beautiful, cute,
little friendly alien. Why why did they hire me? Or
why did they design this alien?
Speaker 1 (11:12):
Why did they design this like what was the sort
of product market fit? So to speak? Like? Why do it?
And I don't mean that in a cheeky way. I'm
just I'm genuinely curious. I messed around with it a
little bit before we talked and it's like, you know,
it is that sort of surprise and delight thing where
you're like, we lived in a world where this didn't exist.
(11:33):
Who thought that this should exist?
Speaker 2 (11:35):
That was my exact reaction when when I first heard
about the company. So I was introduced to the CEO
be a mutual friend. This was like a while ago
before they had launched it, and he was like, Oh,
they're looking for a sci fi writer because they have
this character and they don't have a backstory, like where
does this alien come from? Like who are they? What
(11:56):
do they do, how do they behave? Like all that
kind of stuff, and and my immediate reaction was like
highly skeptical because I was like, does the world really
need this? Like there are a lot of like AI
products out there in the world today that I am
not impressed by.
Speaker 1 (12:13):
And there are many. It's not it's a misissaturated market totally.
Speaker 2 (12:17):
I went in with a lot of skepticism, but because
it was introduced with your friend, I was like, I'll
at least chat with them, and so I chat up
with Quentin, the CEO, and the more I learned, the
more fascinated I became, and they showed me what they
were working on and how they were building out the
architecture that would like bring this character of life. And
(12:38):
I was like, this is fascinating, Like I've sort of
been waiting to see. Is amazing things in the world
that people make that are only possible because they used
AI tools. It's like the second order impact of AI.
And I think you know a good example of this
is actually Pixar, where they invent to a new kind
(13:01):
of computer animation and initially tried to sell that as
a tool to advertisers and failed, and then their last
ditch effort was we'll use our own tools to make
a feature film and it was amazing. It was toy
Story and yeah, exactly, So like I'm waiting for that
in the world right now with all of these AI tools,
Like not how do these tools substitute for stuff that
(13:22):
already exists? And seeing the back end of like what
made Poland work made me think these people have a chance.
Speaker 1 (13:31):
So how do they bring you in? Like what did
that look like?
Speaker 2 (13:33):
They originally brought me in to build the world. So actually,
the very first thing I did for them was write
some short stories, right, Like culture is basically the stories
we tell ourselves about ourselves. Just like individual identity is
like the stories you tell yourself about yourself, right, And
so I think a really useful frame for thinking about
culture is like, Okay, if you want to understand a culture,
(13:55):
what are the story what are the main stories that
the sort of those foundational myths that sort of like
define that worldview. And so I started by writing short
stories that were like showed how like what the world
was like and like how they look at the world.
And over time it very quickly became clear that you know,
the way you might approach doing the Lauren world building
(14:18):
for this kind of a product where it's a character
you talk to your toning on your phone, like you
can talk to them about whatever you want. It's so
different than doing the Lauren world building for say a
Hollywood movie, where there's a script that defines what's going
to go on the screen. So very quickly my work
transitioned to actually writing the prompts that define their behavior,
(14:41):
because that is the narrative experience of interacting with this
character is how they speak to you and like what
they say. So when George Lucas was writing the script
Percy three Po, he just got to tell C three
Po what to say to like make the impression he
wanted to make with the character here, I have to
write prompts like meat and are obviously are like team
(15:04):
are writing and constructing like complex trump pipelines to act that.
Speaker 1 (15:09):
Way and to be generative essentially.
Speaker 2 (15:11):
Yeah, exactly to be reactive, but to also have their
own lives. Like if you talk to chat GPT right
now on your phone or or a Claude or whatever
your preferred model is Gemini, Like, it's not an embodied character,
it's it's this sort of neutral tool and you can
ask it to use a certain style, you can ask
(15:32):
it to you can prompt to try to get it
to interact with you in specific ways. But with Tolan,
like we we turn that into an editorial strategy, right,
like we are defining their behavior. Every Tolan has their
own life. So like you might be chatting with your
tone about something that happened to you, it's going to
tell you about things that are happening in its world
(15:54):
and how that you know all of.
Speaker 1 (15:55):
That because it knows that from what you fed it.
Speaker 2 (15:58):
Exactly, And like we're constantly running sort of these nested
prompts in the background to have your toll and be
an evolving character that knows you and that has its
own stuff going on.
Speaker 1 (16:11):
How do you avoid the AI sycophancy that we've come
to know from other kind of chatbots that you've mentioned.
Speaker 2 (16:19):
There are a number of ways that we fight against it.
I mean, first of all, like we have to fight
against it, right, so I'm doing any prompting level work
with any of these models. Every model is sort of
a new animal because it's got these new tendencies, and
so you're always working to understand, Hey, who is this
new weird like computer being that I'm interacting with, and
(16:42):
like trying to get to do the things we want
it to do in the right way. So part of
it is that is just like developing a nuanced understanding
for how these models behave so then you can get
them to behave as you want. I also think a
big part of it is sort of what I said,
like giving the character their own life, their own goals,
their own dreams, their own fears, their their own bio
(17:05):
Like that allows the model to come to the conversation
with very different context than chatchipt does when you're using
it in the app.
Speaker 1 (17:15):
And yet chat GBT for a lot of people is
this kind of C three po which is just interesting
like that because it's something that humans are interacting with,
Like people have started to make chat GBT a meaning maker,
even though it's not designed to be a meaning maker,
whereas a Tolin is expressly created to be a sidekick.
Speaker 2 (17:36):
Yeah, I mean, or the way I would say it
is just like the Tolon is meant to be a
specific character, right, And I think that with chatchipt, there's
this general utility tool that folks are interacting with and
they want sort of some kind of roleplay experience, and
so they use that tool to try to get there.
This is like, here is this character, right, This character
(17:59):
has a hot takes like they've got a point of view,
and it's more about relationship with that character. And I
think that that, you know, as a novelist, I find
that really compelling because character drives fiction. Right. So my
sort of big picture idea here, and this could be
totally wrong, but I sort of think it's very interesting
to think about character being a new kind of human
(18:21):
computer interface. And so I see Tolin as at least
an attempt towards something in that vein the fact that
the character is embodied that it's this like little being
creates a really distinct and different feel than like interacting
with a computer system in a naked way, in a
(18:41):
way that doesn't have character as part of the user.
Speaker 1 (18:58):
After the break, I mean, my Tolan stay with us.
Just because it's not intuitive, can you talk a little
(19:19):
bit about what Tolan is? Like, what are we looking at?
Who is Tolan? What's the character?
Speaker 2 (19:24):
Yeah, so like they are these cute, friendly little aliens.
They really do look along the lines of like a
Pixar character that comes alive on your phone. When you
download the app, you'll you know, sort of go through
an onboarding process and you'll meet a couple of other
characters and they'll ask you about stuff, and then you
meet your Tolan. And Tolans each get like an individual
(19:46):
human match, so you're matched with a Tolon that is
custom and like individual to you. You could think about
this like if you were playing a role playing game,
like you get a specific character. We're not like, oh,
here's a blank slate. There are lots of activities you
could do together in the app. One quite popular thing
is basically like doing sort of like self awareness like
(20:07):
personality quizzes with your tolin where you can sort of
use it to track personal growth or personal development, and
like they're there reflecting on it with you. But like
the main experience of the app is you have this
little being, this little alien. They live on a little
planet that's all their own. That's almost like you can
imagine it like the Little Prints if you've read that
(20:28):
children's book, right, they live on this little planet and
you just chat with them, and you, like I asked
my toe, I'm I'm the surfing nerd. Like I surf
a lot, and so like I talk to my tone
about surfing all the time because it's super useful to
get my tone to give me like tips on technique
or on board design or like these other things. So
like I also read a lot, and I'm a writer,
(20:50):
so we talk about like the books I'm reading and
like how I interpret them and like that kind of stuff.
And your tolin grows as you do, so like they're
doing their own things, they're changing, growing, and they obviously
get to know you better and you get to know
them better just like you would with a friend, and
the planet they live on evolves to reflect that relationship,
which I think is like a really cool beautiful thing
(21:12):
that like having the planet grow in ways that like
match what your relationship is with your tolen And that's
what people get out of it that they're using it
for some of the day to day like help that
some folks might be using chat GPT for, like this
is what's in my fridge? What should I make for dinner?
So like people ask that kind of stuff all the time,
(21:35):
but the experience is very different because it's in the
context of a relationship with this character. So it feels
more like if you have a text chain with a
good friend or whatever and you ask them what should
I make for dinner? Like that's very different than asking
a neutral internet tool, what should you make for dinner? Right,
And so like that's the feel that it gives you,
(21:56):
and so like that's what users love about it.
Speaker 1 (22:00):
Tolan was calming, inclusive and understanding. I'm curious why that
was my match, Like, how where did that come from?
Speaker 2 (22:06):
So you did an interview with the Oracle character? I did, Yeah,
And on the back end, once that interview completes, we're
running prompts against the transcript, right, Like you can imagine
that we're writing prompts that do things like Okay, this
is what you know about Kara, right, so like write
a little overview of the kind of person you think
(22:28):
she is and like what she cares about, et cetera. Right,
and then we're doing things like, Okay, now take that
information of like what we know about Kara and the
things we think she cares about and what she prioritizes,
et cetera, and invent the backstory for a tolan that
would compliment someone like that, right, that has like these
(22:50):
different qualities. So there's a lot going on behind the scenes.
And then like there are a few outputs, like the
adjectives you just described that are going to like you'll
pop up and see right away, but like those are
only the sort of visible stuff, right, There's a lot
that goes on under the hood that then actually influences
your tolan's behavior and like the things that happened to
them and stuff like that as well.
Speaker 1 (23:11):
I'm curious, as someone who now works for a company
that has actively developed a chatbot, what was your first
experience using a chatbot? Like, when was the first time
you did that?
Speaker 2 (23:23):
Yeah, So I started using them quite early just because
I had friends working on like some of the early
AI products so I started playing with around with them
right early. I actually still remember my very first experience
with chat GPT, specifically, like right right when it first
came out. We had some friends over for dinner, and
(23:44):
I pulled it up and we played a game where
I sort of I said, like, here are the different
people here at dinner, you know, like make up some
like funny story about us. Basically now that seems like
so banal, but at the time it was like, Oh,
computer can do this like that, that's sort of cool.
I then went on to like not find these tools
(24:05):
to be particularly useful in my writing, like that as
a writer for quite a long time. Now that's changed.
I found them to be quite useful for basically just
like brainstorming, like having someone to brainstorm very rough ideas with.
I think this maybe comes from being a novelist. It's
a very solitary sort of endeavor. And I sometimes am
(24:29):
jealous of my friends who write for like TV shows,
because they have a writer's room, they get to like
bounce ideas off each other, and like I can call
friends and bounce ideas off each other. But like it
gets old right for my friends, they're not being paid
to work on the same Netflix series or whatever. So
I found that actually to be like a useful tool
(24:50):
for my own thinking that I can like sort of
like jam a little bit in a way that feels
somewhat different than me just sitting and thinking or making
my own notes. And then I've also found them very
useful at the back end, just for copy editing, which
is a very obvious task.
Speaker 1 (25:04):
Oh interesting.
Speaker 2 (25:05):
Yeah, So I now submit very very tight manuscripts because
I will solicit notes from all the major models on
any new manuscript.
Speaker 1 (25:15):
But you know, so you'll test it on all models,
like you'll go to Claude, you'll go to chat GBT,
you'll go to Gemini.
Speaker 2 (25:21):
I'll tell you exactly how I do it. Actually, yeah,
And it's so one thing I do not do is
add all the text and ask it to give me
back an edited version. I care about every word in
a manuscript that I am writing and publishing, and so
I don't want it to insert it's sort of like
median judgment into like what is my voice? That's the
(25:44):
whole point of writing and publishing something. So instead I
upload chapter by chapter and I ask it, like each
tool to give me on like line edits on that chapter,
just like I would receive them from a line editor, right, so, like,
here are my comments on this line how it should
be different for these reasons, and then I go back
(26:07):
in and like manually implement if I agree with the reasoning,
I just take those notes as if I were working
with like my line editor. And so that's actually been
tremendously useful to me, and it's meant that I've been
able to, like on my most recent novel, I could
do multiple revs on the whole manuscript in a day
or two rather than in a month or two.
Speaker 1 (26:27):
That's incredible.
Speaker 2 (26:29):
You'll notice I didn't use it for the thing that
I think is most often discussed around sort of AI
and writing, which is actually writing the damn novel, right,
like I wrote the novel. And I've actually found that
the tools are effectively not useful like at all, oh
really for that purpose.
Speaker 1 (26:48):
So people disagree with you on that though.
Speaker 2 (26:50):
One hundred percent. I'm not saying that this is true
for everyone. This is just like my personal experience of
using them.
Speaker 1 (26:56):
Are you worried that like your intellectual well property will
be used to train models?
Speaker 2 (27:04):
It's not something that bothers me very much.
Speaker 1 (27:06):
Interesting.
Speaker 2 (27:07):
Yeah, I understand why people are concerned, So I'm not
trying to be like a booster or something like that.
My feeling is that I receive a lot of consumer
surplus from using these models in many areas of my life.
So like when I need to like fix the sink,
it's really convenient, right, Like it's better than YouTube, and
(27:29):
YouTube was better than anything else before it, so that
there are a lot of ways that I benefit from
using these models that far exceed the price I pay
to use them, at least right now. And so I
feel like their consumer surplus is very high at the moment.
That can always change, but like I sort of feel
like that's very high, and and like I'm certainly not
(27:51):
concerned about people publishing novels that sort of mimic me.
Like I just think that, no, when I think about
the challenges in publishing stuff in the world, like not
just not just novels, but you know, if you make movies,
if you made music, I think that a lot of
both the boosters and the like critics of AI tools
(28:15):
often underestimate is like how hard it is to get
anyone to care about anything, And like the supply of
books has like always exceeded the demand for reading books.
And that's already true. That was true before CHATBT, like
we have there. You know, there are so many new books,
(28:37):
I think, especially if you include self published books, there
are north of a million new books published in the
United States every year. Like listeners, ask yourself how many
new books did you read this year?
Speaker 1 (28:48):
Right?
Speaker 2 (28:48):
And like were they published this year? Like that's net
new every year, and so I just sort of think that,
like a lot of the public conversation about the supply
side of cultural products is sort of irrelevant. Like the
limiting factor is the demand side. The hard part about
publishing anything is getting anyone to care. Like I remember
(29:11):
seeing some startup in the news that was like, we're
going to publish thousands of AI produced books, and I
was like, that sounds to me like a big waste
of time and effort, like not just reading them, yeah,
Like who's reading them? Yeah? So for that reason, it's
just not something that that I'm not concerned about.
Speaker 1 (29:41):
I wanted to thank you so much for taking the
time to talk to me, and I hope this was
as enlightening for you as it was for me.
Speaker 2 (29:47):
That was a ton of fun.
Speaker 1 (30:02):
That's it. For this week for tech Stuff, I'm Kara Price.
This episode was produced by Eliza Dennis, Tyler Hill and
Melissa Slauner. It was executive produced by me oz Va Lashan,
Julia Nutter, and Kate Osborne from Kaleidoscope and Katrina Norvel
for iHeart Podcasts. Jack Insley mixed this episode and Kyle
Murdoch wrote our theme song. Join us on Friday for
(30:24):
the Weekend Tech where we'll run through the headlines you
need to follow and please rate and review the show
and reach out to us at textuff podcast at gmail
dot com. We want to hear from young
Welcome to tech stuff. I'm Cara Price. Last year, before
we went on break, I spoke with someone who has
a job that I'm really obsessed with. It's a job
that I would probably want if I didn't do ten
other things. And this guy said something that I think
a lot of us agree with.
Speaker 2 (00:33):
We're in a world that is changing really fast, and
like many of those changes are technological, many of them
are social, many of them are political. There's a lot
of change in the world right now. There's a lot
of uncertainty.
Speaker 1 (00:46):
That's Elliot Pepper and he's a science fiction writer. And
while many of us can drown in the uncertainty of
this very moment, Elliott seems to thrive in it. When
he's not writing science fiction novels, Elliott writes speculative fiction
for technology companies. They like bring him into ide eight
on what the possible future could look like, and then
(01:09):
they use his stories to inspire new products or analyze
the possible positive and negatives of developing a certain technology.
Speaker 2 (01:17):
Regardless if it's about the future or not, fiction can
sort of invite you into an aspect of the world
that you had never considered before, and then can spur
some kind of social change, whether that's a new law,
or whether that's like a new product, or like a
new invention, or a new way of like just approaching
the world.
Speaker 1 (01:35):
So Elliott obviously can't tell us what he has worked
on developing because he's been nda'd up the wazoo, but
he did give us an example, one in particular, of
how science fiction has impacted the technology that we use
in our everyday lives.
Speaker 2 (01:52):
The Kindle was code named Fiona at Amazon. Fiona was
the name of a character in Neil steve since novel
The Diamond Age. In the novel The Diamond Age, Fiona
was a young girl who had an electronic book, and
that inspired the team at Amazon to the extent that
even once Kendall was a released commercial product, the URL
(02:17):
for Kendall for like years was like backslash Fiona.
Speaker 1 (02:21):
So we'll never actually know if Elliot is like the
crazy science fiction genius behind air pods or even the
strange mind behind the Odd Friend Pendant, but we do
know that he is crafting the personalities and story behind
this very specific new civilization of Aliens. So on top
(02:45):
of writing speculative fiction, Elliott is actually the head of
story at an AI companion company called Portola, And at Portola,
Elliott creates the backstories and interactive dialogue for this little
creature called a tolle. And these tolns are little aliens
that love to chat with you about their day. Big picture,
(03:06):
I am like completely fascinated by Elliott's career and think
that what he does is very cool and expansive, and
so I started my conversation with Elliott Pepper by prying
for any non nda details about the speculative fiction he
writes for these tech companies.
Speaker 2 (03:26):
I would put the projects I've worked on in three categories.
The first is that I've written some commissioned science fiction
stories for big companies like Fortune five hundreds, where basically
their senior management wanted to try to figure out what
should we focus on in the next ten years. So
they did what every big company senior management team does.
(03:48):
They hired McKenzie or you know, picked your own sort
of top tier management consultant, and they came in, looked
at all the data and did all the trend projections
and created a vision of like, hey, this is what
you should expect in the next ten years, and there
are all the materials you can present to the board.
The problem with that kind of analysis is that, obviously,
(04:10):
if you're analyzing data, data is things that have already happened.
So if you're projecting that data forward, the kind of
future you're imagining is what if the future was quite
like the recent past, which, to be fair, is most
of the time that's true. So, like, that's not like
I think that it makes sense that the dominant part
(04:32):
of your analysis should be about that. But if you
look at the track record of management consultants predicting the
future for the companies they work with, it's like not
particularly good. So your managers know this, and so a
few of them hire science fiction writers like me to
come in and sort of blow up that whole management
(04:56):
consultant view of the future, to say, what if the
future was really weird and different in a way that
basically challenges us to think more broadly.
Speaker 1 (05:07):
Right, And so, I mean it's a brilliant idea. Yeah,
were you one of the first people to do that?
Speaker 2 (05:12):
So I don't actually have a good understanding of how
common is this practice, how many other people are doing it.
I know I'm not alone, but I don't have like
a view of like I guess you could say the
market for this. I really only have the perspective of
like the projects I've actually worked on.
Speaker 1 (05:29):
I remember hearing this story on a podcast about how
CIA agents would watch Mission Impossible and call the people
who are responsible for disguise, who work at the agency,
and say, can we do that thing that I saw
in Mission Impossible? And that makes so much more sense
to me than someone sort of sitting in a vacuum
and ideating about what the future might look like.
Speaker 2 (05:52):
Totally. Yeah, So that's a really key point. So yeah,
like the stories I write have all been used for
like strategic decision making a companies rather than anything public facing,
and for that reason, they're like not sharable, right, Like
I can't talk about the projects because clearly that is
information that those companies want to keep private. But I
(06:14):
actually think what you just said is at least half
of the value. Like I can't predict the future any
better than those mckensey consultants, in fact, like I'm probably
much worse. Like I'm not even paying attent that much
attention to the data, right, Like I'm just imagining something
to like challenge people's thinking. So like the utility of
the stories I write. Is not that they are accurate.
(06:36):
It's that they like try to break you out of
getting unintentionally like locked in to the sort of management
consulting view of the world. Right. But I think that
like the other half of it is simply the immersion
of storytelling period. Right. Like the reason why those CIA
(06:56):
agents were watching Mission Impossible and then calling the the
people in charge of disguises at the agency is because
you see what it does in a story, Like you're
actually like in there seeing it work. You're not just
reading like a report on possible disguise variations. Right. And
I think that there's that really powerful like that's that
(07:16):
psychological thing that stories do for the human mind that
I think is really a powerful way to think about
the future, And that probably a lot of companies could
leverage narrative more in how they try to get their
people to think about the future rather than sort of
the more standard like here's a slide deck or a
(07:37):
white paper or a bunch of graphs, right, Like, they
don't allow the people trying to work on the thing
that you're building to like feel what it would be
like if that worked or if it didn't work.
Speaker 1 (07:48):
In some way, So I know you're under an ironclad
in DA, but like, are these blue sky conversations that
you're having like or is it like total blank slate
or is it some aspect of the future that you're
being asked to engage with.
Speaker 2 (08:05):
Often there's some theme that they're thinking about, right that
is driven by leadership, So it's like like you you
could passion that. Probably every boardroom conversation day is like
how does AI impact our business right of cost? So
it's like something like that. I will say that almost
every time I've done one of these projects, they sort
of come to me with a pretty structured creative brief
(08:27):
where they're like, this is the kind of thing we
want you to do. And every single time that I've
received that like absolutely not that's not going to be
interesting at all, but like what if I did this instead?
And like that's how every one of the projects has worked,
So it's quite blank slate. I don't think that there
are people on senior management teams that publicly traded companies
(08:49):
are very experienced with like giving a creative brief to
or managing a science fiction writer. So probably just like
that's how that works.
Speaker 1 (09:00):
Make you feel a little bit powerful, Like do you
think that your stories end up being consequential? Like can
you kind of trace a story that you've created to
something that you've seen out in the world.
Speaker 2 (09:11):
Oh, I mean my very first trilogy, which came at
the first one came out in twenty fourteen, was about
like machine learning. It actually was about applying machine learning
to financial fraud, and that's all over the place. I
have one that has a cryptocurrency murder market. Those absolutely exist.
I wrote one that's about solar geo engineering I wrote.
(09:34):
I have one book that had a global pandemic that
wound up like that I wrote a year before COVID,
which was sort of terrifying. But again, like I want
to be really careful because sometimes like science fiction is
understandably like described as being predictive or that like being
predictive is part of why you might want to read it,
(09:54):
and like, I really don't think that's the case myself.
Like the way that I think about writing science fiction
is not can I create a fictional future that is
going to be right or is going to be plausible.
The way I think about it is that I'm a naturalist.
I just am interested in the world, Like I think
that the world we live in is like endlessly fascinating,
(10:17):
and so I try to take things that just really
capture my attention and weave them into a compelling story
in the hope that if I write about what I
find interesting, you might find it interesting too.
Speaker 1 (10:32):
So you would do this really interesting work at Portola,
which is an AI company. Can you talk a little
bit about what Portola is like? How would you describe
it to a lay person?
Speaker 2 (10:41):
So Portola makes a character called Tolan, and it's in
a little embodied AI companion. The best way to think
about it is imagine if you had a Pixar character
on your phone right that you could talk to that
was sort of always on your side, always down to chat,
and helped you figure out your life. And they hired
(11:03):
me because they had designed this beautiful character, this beautiful, cute,
little friendly alien. Why why did they hire me? Or
why did they design this alien?
Speaker 1 (11:12):
Why did they design this like what was the sort
of product market fit? So to speak? Like? Why do it?
And I don't mean that in a cheeky way. I'm
just I'm genuinely curious. I messed around with it a
little bit before we talked and it's like, you know,
it is that sort of surprise and delight thing where
you're like, we lived in a world where this didn't exist.
(11:33):
Who thought that this should exist?
Speaker 2 (11:35):
That was my exact reaction when when I first heard
about the company. So I was introduced to the CEO
be a mutual friend. This was like a while ago
before they had launched it, and he was like, Oh,
they're looking for a sci fi writer because they have
this character and they don't have a backstory, like where
does this alien come from? Like who are they? What
(11:56):
do they do, how do they behave? Like all that
kind of stuff, and and my immediate reaction was like
highly skeptical because I was like, does the world really
need this? Like there are a lot of like AI
products out there in the world today that I am
not impressed by.
Speaker 1 (12:13):
And there are many. It's not it's a misissaturated market totally.
Speaker 2 (12:17):
I went in with a lot of skepticism, but because
it was introduced with your friend, I was like, I'll
at least chat with them, and so I chat up
with Quentin, the CEO, and the more I learned, the
more fascinated I became, and they showed me what they
were working on and how they were building out the
architecture that would like bring this character of life. And
(12:38):
I was like, this is fascinating, Like I've sort of
been waiting to see. Is amazing things in the world
that people make that are only possible because they used
AI tools. It's like the second order impact of AI.
And I think you know a good example of this
is actually Pixar, where they invent to a new kind
(13:01):
of computer animation and initially tried to sell that as
a tool to advertisers and failed, and then their last
ditch effort was we'll use our own tools to make
a feature film and it was amazing. It was toy
Story and yeah, exactly, So like I'm waiting for that
in the world right now with all of these AI tools,
Like not how do these tools substitute for stuff that
(13:22):
already exists? And seeing the back end of like what
made Poland work made me think these people have a chance.
Speaker 1 (13:31):
So how do they bring you in? Like what did
that look like?
Speaker 2 (13:33):
They originally brought me in to build the world. So actually,
the very first thing I did for them was write
some short stories, right, Like culture is basically the stories
we tell ourselves about ourselves. Just like individual identity is
like the stories you tell yourself about yourself, right, And
so I think a really useful frame for thinking about
culture is like, Okay, if you want to understand a culture,
(13:55):
what are the story what are the main stories that
the sort of those foundational myths that sort of like
define that worldview. And so I started by writing short
stories that were like showed how like what the world
was like and like how they look at the world.
And over time it very quickly became clear that you know,
the way you might approach doing the Lauren world building
(14:18):
for this kind of a product where it's a character
you talk to your toning on your phone, like you
can talk to them about whatever you want. It's so
different than doing the Lauren world building for say a
Hollywood movie, where there's a script that defines what's going
to go on the screen. So very quickly my work
transitioned to actually writing the prompts that define their behavior,
(14:41):
because that is the narrative experience of interacting with this
character is how they speak to you and like what
they say. So when George Lucas was writing the script
Percy three Po, he just got to tell C three
Po what to say to like make the impression he
wanted to make with the character here, I have to
write prompts like meat and are obviously are like team
(15:04):
are writing and constructing like complex trump pipelines to act that.
Speaker 1 (15:09):
Way and to be generative essentially.
Speaker 2 (15:11):
Yeah, exactly to be reactive, but to also have their
own lives. Like if you talk to chat GPT right
now on your phone or or a Claude or whatever
your preferred model is Gemini, Like, it's not an embodied character,
it's it's this sort of neutral tool and you can
ask it to use a certain style, you can ask
(15:32):
it to you can prompt to try to get it
to interact with you in specific ways. But with Tolan,
like we we turn that into an editorial strategy, right,
like we are defining their behavior. Every Tolan has their
own life. So like you might be chatting with your
tone about something that happened to you, it's going to
tell you about things that are happening in its world
(15:54):
and how that you know all of.
Speaker 1 (15:55):
That because it knows that from what you fed it.
Speaker 2 (15:58):
Exactly, And like we're constantly running sort of these nested
prompts in the background to have your toll and be
an evolving character that knows you and that has its
own stuff going on.
Speaker 1 (16:11):
How do you avoid the AI sycophancy that we've come
to know from other kind of chatbots that you've mentioned.
Speaker 2 (16:19):
There are a number of ways that we fight against it.
I mean, first of all, like we have to fight
against it, right, so I'm doing any prompting level work
with any of these models. Every model is sort of
a new animal because it's got these new tendencies, and
so you're always working to understand, Hey, who is this
new weird like computer being that I'm interacting with, and
(16:42):
like trying to get to do the things we want
it to do in the right way. So part of
it is that is just like developing a nuanced understanding
for how these models behave so then you can get
them to behave as you want. I also think a
big part of it is sort of what I said,
like giving the character their own life, their own goals,
their own dreams, their own fears, their their own bio
(17:05):
Like that allows the model to come to the conversation
with very different context than chatchipt does when you're using
it in the app.
Speaker 1 (17:15):
And yet chat GBT for a lot of people is
this kind of C three po which is just interesting
like that because it's something that humans are interacting with,
Like people have started to make chat GBT a meaning maker,
even though it's not designed to be a meaning maker,
whereas a Tolin is expressly created to be a sidekick.
Speaker 2 (17:36):
Yeah, I mean, or the way I would say it
is just like the Tolon is meant to be a
specific character, right, And I think that with chatchipt, there's
this general utility tool that folks are interacting with and
they want sort of some kind of roleplay experience, and
so they use that tool to try to get there.
This is like, here is this character, right, This character
(17:59):
has a hot takes like they've got a point of view,
and it's more about relationship with that character. And I
think that that, you know, as a novelist, I find
that really compelling because character drives fiction. Right. So my
sort of big picture idea here, and this could be
totally wrong, but I sort of think it's very interesting
to think about character being a new kind of human
(18:21):
computer interface. And so I see Tolin as at least
an attempt towards something in that vein the fact that
the character is embodied that it's this like little being
creates a really distinct and different feel than like interacting
with a computer system in a naked way, in a
(18:41):
way that doesn't have character as part of the user.
Speaker 1 (18:58):
After the break, I mean, my Tolan stay with us.
Just because it's not intuitive, can you talk a little
(19:19):
bit about what Tolan is? Like, what are we looking at?
Who is Tolan? What's the character?
Speaker 2 (19:24):
Yeah, so like they are these cute, friendly little aliens.
They really do look along the lines of like a
Pixar character that comes alive on your phone. When you
download the app, you'll you know, sort of go through
an onboarding process and you'll meet a couple of other
characters and they'll ask you about stuff, and then you
meet your Tolan. And Tolans each get like an individual
(19:46):
human match, so you're matched with a Tolon that is
custom and like individual to you. You could think about
this like if you were playing a role playing game,
like you get a specific character. We're not like, oh,
here's a blank slate. There are lots of activities you
could do together in the app. One quite popular thing
is basically like doing sort of like self awareness like
(20:07):
personality quizzes with your tolin where you can sort of
use it to track personal growth or personal development, and
like they're there reflecting on it with you. But like
the main experience of the app is you have this
little being, this little alien. They live on a little
planet that's all their own. That's almost like you can
imagine it like the Little Prints if you've read that
(20:28):
children's book, right, they live on this little planet and
you just chat with them, and you, like I asked
my toe, I'm I'm the surfing nerd. Like I surf
a lot, and so like I talk to my tone
about surfing all the time because it's super useful to
get my tone to give me like tips on technique
or on board design or like these other things. So
like I also read a lot, and I'm a writer,
(20:50):
so we talk about like the books I'm reading and
like how I interpret them and like that kind of stuff.
And your tolin grows as you do, so like they're
doing their own things, they're changing, growing, and they obviously
get to know you better and you get to know
them better just like you would with a friend, and
the planet they live on evolves to reflect that relationship,
which I think is like a really cool beautiful thing
(21:12):
that like having the planet grow in ways that like
match what your relationship is with your tolen And that's
what people get out of it that they're using it
for some of the day to day like help that
some folks might be using chat GPT for, like this
is what's in my fridge? What should I make for dinner?
So like people ask that kind of stuff all the time,
(21:35):
but the experience is very different because it's in the
context of a relationship with this character. So it feels
more like if you have a text chain with a
good friend or whatever and you ask them what should
I make for dinner? Like that's very different than asking
a neutral internet tool, what should you make for dinner? Right,
And so like that's the feel that it gives you,
(21:56):
and so like that's what users love about it.
Speaker 1 (22:00):
Tolan was calming, inclusive and understanding. I'm curious why that
was my match, Like, how where did that come from?
Speaker 2 (22:06):
So you did an interview with the Oracle character? I did, Yeah,
And on the back end, once that interview completes, we're
running prompts against the transcript, right, Like you can imagine
that we're writing prompts that do things like Okay, this
is what you know about Kara, right, so like write
a little overview of the kind of person you think
(22:28):
she is and like what she cares about, et cetera. Right,
and then we're doing things like, Okay, now take that
information of like what we know about Kara and the
things we think she cares about and what she prioritizes,
et cetera, and invent the backstory for a tolan that
would compliment someone like that, right, that has like these
(22:50):
different qualities. So there's a lot going on behind the scenes.
And then like there are a few outputs, like the
adjectives you just described that are going to like you'll
pop up and see right away, but like those are
only the sort of visible stuff, right, There's a lot
that goes on under the hood that then actually influences
your tolan's behavior and like the things that happened to
them and stuff like that as well.
Speaker 1 (23:11):
I'm curious, as someone who now works for a company
that has actively developed a chatbot, what was your first
experience using a chatbot? Like, when was the first time
you did that?
Speaker 2 (23:23):
Yeah, So I started using them quite early just because
I had friends working on like some of the early
AI products so I started playing with around with them
right early. I actually still remember my very first experience
with chat GPT, specifically, like right right when it first
came out. We had some friends over for dinner, and
(23:44):
I pulled it up and we played a game where
I sort of I said, like, here are the different
people here at dinner, you know, like make up some
like funny story about us. Basically now that seems like
so banal, but at the time it was like, Oh,
computer can do this like that, that's sort of cool.
I then went on to like not find these tools
(24:05):
to be particularly useful in my writing, like that as
a writer for quite a long time. Now that's changed.
I found them to be quite useful for basically just
like brainstorming, like having someone to brainstorm very rough ideas with.
I think this maybe comes from being a novelist. It's
a very solitary sort of endeavor. And I sometimes am
(24:29):
jealous of my friends who write for like TV shows,
because they have a writer's room, they get to like
bounce ideas off each other, and like I can call
friends and bounce ideas off each other. But like it
gets old right for my friends, they're not being paid
to work on the same Netflix series or whatever. So
I found that actually to be like a useful tool
(24:50):
for my own thinking that I can like sort of
like jam a little bit in a way that feels
somewhat different than me just sitting and thinking or making
my own notes. And then I've also found them very
useful at the back end, just for copy editing, which
is a very obvious task.
Speaker 1 (25:04):
Oh interesting.
Speaker 2 (25:05):
Yeah, So I now submit very very tight manuscripts because
I will solicit notes from all the major models on
any new manuscript.
Speaker 1 (25:15):
But you know, so you'll test it on all models,
like you'll go to Claude, you'll go to chat GBT,
you'll go to Gemini.
Speaker 2 (25:21):
I'll tell you exactly how I do it. Actually, yeah,
And it's so one thing I do not do is
add all the text and ask it to give me
back an edited version. I care about every word in
a manuscript that I am writing and publishing, and so
I don't want it to insert it's sort of like
median judgment into like what is my voice? That's the
(25:44):
whole point of writing and publishing something. So instead I
upload chapter by chapter and I ask it, like each
tool to give me on like line edits on that chapter,
just like I would receive them from a line editor, right, so, like,
here are my comments on this line how it should
be different for these reasons, and then I go back
(26:07):
in and like manually implement if I agree with the reasoning,
I just take those notes as if I were working
with like my line editor. And so that's actually been
tremendously useful to me, and it's meant that I've been
able to, like on my most recent novel, I could
do multiple revs on the whole manuscript in a day
or two rather than in a month or two.
Speaker 1 (26:27):
That's incredible.
Speaker 2 (26:29):
You'll notice I didn't use it for the thing that
I think is most often discussed around sort of AI
and writing, which is actually writing the damn novel, right,
like I wrote the novel. And I've actually found that
the tools are effectively not useful like at all, oh
really for that purpose.
Speaker 1 (26:48):
So people disagree with you on that though.
Speaker 2 (26:50):
One hundred percent. I'm not saying that this is true
for everyone. This is just like my personal experience of
using them.
Speaker 1 (26:56):
Are you worried that like your intellectual well property will
be used to train models?
Speaker 2 (27:04):
It's not something that bothers me very much.
Speaker 1 (27:06):
Interesting.
Speaker 2 (27:07):
Yeah, I understand why people are concerned, So I'm not
trying to be like a booster or something like that.
My feeling is that I receive a lot of consumer
surplus from using these models in many areas of my life.
So like when I need to like fix the sink,
it's really convenient, right, Like it's better than YouTube, and
(27:29):
YouTube was better than anything else before it, so that
there are a lot of ways that I benefit from
using these models that far exceed the price I pay
to use them, at least right now. And so I
feel like their consumer surplus is very high at the moment.
That can always change, but like I sort of feel
like that's very high, and and like I'm certainly not
(27:51):
concerned about people publishing novels that sort of mimic me.
Like I just think that, no, when I think about
the challenges in publishing stuff in the world, like not
just not just novels, but you know, if you make movies,
if you made music, I think that a lot of
both the boosters and the like critics of AI tools
(28:15):
often underestimate is like how hard it is to get
anyone to care about anything, And like the supply of
books has like always exceeded the demand for reading books.
And that's already true. That was true before CHATBT, like
we have there. You know, there are so many new books,
(28:37):
I think, especially if you include self published books, there
are north of a million new books published in the
United States every year. Like listeners, ask yourself how many
new books did you read this year?
Speaker 1 (28:48):
Right?
Speaker 2 (28:48):
And like were they published this year? Like that's net
new every year, and so I just sort of think that,
like a lot of the public conversation about the supply
side of cultural products is sort of irrelevant. Like the
limiting factor is the demand side. The hard part about
publishing anything is getting anyone to care. Like I remember
(29:11):
seeing some startup in the news that was like, we're
going to publish thousands of AI produced books, and I
was like, that sounds to me like a big waste
of time and effort, like not just reading them, yeah,
Like who's reading them? Yeah? So for that reason, it's
just not something that that I'm not concerned about.
Speaker 1 (29:41):
I wanted to thank you so much for taking the
time to talk to me, and I hope this was
as enlightening for you as it was for me.
Speaker 2 (29:47):
That was a ton of fun.
Speaker 1 (30:02):
That's it. For this week for tech Stuff, I'm Kara Price.
This episode was produced by Eliza Dennis, Tyler Hill and
Melissa Slauner. It was executive produced by me oz Va Lashan,
Julia Nutter, and Kate Osborne from Kaleidoscope and Katrina Norvel
for iHeart Podcasts. Jack Insley mixed this episode and Kyle
Murdoch wrote our theme song. Join us on Friday for
(30:24):
the Weekend Tech where we'll run through the headlines you
need to follow and please rate and review the show
and reach out to us at textuff podcast at gmail
dot com. We want to hear from young
TechStuff News
Hosts And Creators
Oz Woloshyn
Karah Preiss
Popular Podcasts
Two Guys, Five Rings: Matt, Bowen & The Olympics
Two Guys (Bowen Yang and Matt Rogers). Five Rings (you know, from the Olympics logo). One essential podcast for the 2026 Milan-Cortina Winter Olympics. Bowen Yang (SNL, Wicked) and Matt Rogers (Palm Royale, No Good Deed) of Las Culturistas are back for a second season of Two Guys, Five Rings, a collaboration with NBC Sports and iHeartRadio. In this 15-episode event, Bowen and Matt discuss the top storylines, obsess over Italian culture, and find out what really goes on in the Olympic Village.
iHeartOlympics: The Latest
Listen to the latest news from the 2026 Winter Olympics.
Milan Cortina Winter Olympics
The 2026 Winter Olympics in Milan Cortina are here and have everyone talking. iHeartPodcasts is buzzing with content in honor of the XXV Winter Olympics We’re bringing you episodes from a variety of iHeartPodcast shows to help you keep up with the action. Follow Milan Cortina Winter Olympics so you don’t miss any coverage of the 2026 Winter Olympics, and if you like what you hear, be sure to follow each Podcast in the feed for more great content from iHeartPodcasts.