Do you work with machines, or for machines? Cory Doctorow, who coined the term “enshittification,” has a new book that asks exactly that. In The Reverse Centaur's Guide to Life After AI: How to Think About Artificial Intelligence Before It's Too Late, he challenges the belief that AI is inevitable, explains why tech boosters are pushing this message, and how it leads to companies that are “too big to fail… too big to jail… too big to care.”
Cory Doctorow and Oz discuss the economics of a bubble, algorithmic wage discrimination, and what you can do to help save your industry from disruption.
Additional Reading:
See omnystudio.com/listener for privacy information.
Listen
Watch
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:15):
Welcome to tech Stuff. I'm Os Voloshen. If you've been
listening to this podcast, or if you spend any time
on the Internet, you won't be able to miss the
word in shitification. And our guest today is the man
who coined it, Corey doctor Ol. He's a science fiction author,
a technology activist, and a journalist. His new book is
called The Reverse Centaur's Guide to Life After AI, How
(00:39):
to Think about Artificial Intelligence before It's too late. Corey,
Welcome to text Stuff.
Speaker 2 (00:44):
Thanks as it's a pleasure to be on.
Speaker 1 (00:46):
For anyone who did miss it. What is in ghentification?
Speaker 2 (00:49):
Well, I've worked for the Electronic Frontier Foundation, which is
an NGO, for twenty five years on the question digital rights,
and getting people to engage with that question is very
hard because digital rights questions are abstract and technical and
relate to things that are going to happen in the future,
and for very good reasons, people mostly care about concrete
things happening right now. And what I have done over
(01:12):
the years has come up with different framing devices and
similes and metaphors and narratives too. I'm a science fiction novelist,
but it turned out that a dirty word initification was
the way to get people to engage in it, and
so more specifically in sitification is this theory and description.
It describes how the platforms that we rely on have
turned into piles of shit, where first they are good
(01:34):
to their end users but lock them in, and then
having locked in the end users and made it difficult
for them to leave, they make it worse for them
to make it tempting for businesses. And then when the
businesses are also locked in, they also extract everything they
can from those businesses and line their own pockets and
turn into a pile of shit. But the more interesting
part of it is why they're doing it, And this
(01:54):
maybe relates to the conversation we're having today, which is
about the economics and political economy of monopoly, that when
you let firms get too big to fail, they become
too big to jail, and then that makes them too
big to care, and they destroy our lives for the
same reason your dog looks its balls because they can
and no one makes them start.
Speaker 1 (02:13):
I want to get into all of that with you,
but I've read the acknowledgments of your book, not just
the acknowledgements. If I'm always interesting acknowledgments, and you had
the same editor for this book, this new book, the
Reverse Centaur's Guide, as you did for last year's in
Certification book, and you kind of credit him with helping you,
not just with in Certification argument, but with this argument.
(02:34):
And I'm curious, like, where does the reverse central sit
in the lineage of inertifications? What is it a development
of the same argument or is it totally new argument?
Because in Certification was essentially about the tech platforms of
the Internet, and this is about the experience of AI.
Speaker 2 (02:53):
The common lineage between in Certification and my critique of
AI is grounded in the dysfunctions and pathologies of firms
that conquer their markets of saturate their markets. Monopolies, cartels doopolies.
And one of the consequences of having monopoly is that
(03:15):
you become ungovernable in the worst sense, not in the
cool sense of like a raccoon, but in the bad
sense of like Donald Trump. And you are so liberated
from consequence that you can do unlimited bad things to
lots of people, including people you would think would have
some power in the economy, and just get away with it.
(03:36):
And really the worst ideas of the worst people become
the most profitable way of doing business. So that's like
in shitification at its theoretical core. But reverse centaur is
about another epiphenomenon, another thing that a consequence of monopoly,
which is that if your firm that has saturated its
(03:57):
market right Google has a ninety percent market share, then
you can't grow anymore. And you know, there's this saying
among leftists and crunchy granola types that endless growth is
the ideology of a tumor, and that grounds it in
this idea that the reason firms want to continue growing
(04:18):
as ideological and not practical or concrete. And while there
is some ideology at work there in the way that
we think about our economy and in the environmental consequences
of that, there are very practical, material reasons that companies
want to be perceived as growing and not as mature
and at the end of their growth cycle. And that
(04:40):
is because the share price of a company that is
growing is much higher based on the total economic turnover
of that company than the share price of affirm with
the same turnover. But that is static, and that's not irrational.
A share and affirm is a claim on its future earnings.
Firms that are growing have more future earnings to return
(05:03):
to you than firms that are not growing. And the
corollary of this is that when your firm stops growing,
it becomes overvalued because the future earnings have contracted when
you reach the top of your game and you stop growing,
and that provokes these mass sell offs by investors.
Speaker 1 (05:25):
You mentioned, I think you mentioned that you know the
Invidia cell of when the deep Seak moment happened. Obviously, Yeah,
in video rebounded from that, but it raised the it
raised a specter that growth may not be unlimited.
Speaker 2 (05:37):
Yeah, the video sell off was the largest decapitalization in
twenty four hours of any firm. And when your firm
becomes illiquid, right when it's not, when it's not traded
at this high multiple and so liquid, you can't grow
by buying other firms by offering them stock. You have
to give them money. And money is an exogenous substance
that's produced outside the firm, and shares are things that
(05:58):
are produced inside the firm by typing zeros and to
a spreadsheet. And so it's always preferable to be able
to acquire firms and talent using shares. This endogenous substance
you can produce on demand, and you can only do
that while you're growing. So this is another kind of
growth paradox, right, which is that it's easy to keep
growing when you're growing because you can buy other companies,
(06:19):
but it's hard when you stop growing to start growing again,
not least because all those key employees that have been
paid in shares suddenly see their net worthfall off a
cliff when you when your share price declines, and they
might go look for work elsewhere, which means they're not
going to help you start the firm up again. They're
not going to help you get a back on a
growth trajectory.
Speaker 1 (06:37):
So that the central argument of the book is that
a lot of technology companies order the practice order. The
technology companies have an extraordinary incentive to persuade the white
world that in a AI is inevitable, and that that
kind of mythology is of the essence to protecting their
value and therefore keeping their employees and their ability to
(07:00):
keep growing by buying other companies with their stock.
Speaker 2 (07:04):
Yes, So what I would say is that the book
is exploring why tech bubbles exist at all. And this
is not the first one we've had, you know, metaverse
and cryptocurrency and web three, all those other things.
Speaker 1 (07:16):
But they were all palpably confections, right, those those were
not things that people use, whereas, like all you do
is any conversation you over hear in the subway, in
a restaurant, whatever, is people telling each other how they
use AI. So that arguably is a big difference.
Speaker 2 (07:31):
So it's just so those bubbles were smaller. Why is
the a bubble larger? And some of it is because
AI is realer than those things. So I'm a fake
computer scientist. I have an honorary doctor in computer science
from the Open University, and in my capacity as a
fake computer scientist, I can tell you that AI is
pretty interesting.
Speaker 1 (07:50):
And from a user point of view, I would argue
a feeling of technology magic for the first time in
a long time.
Speaker 2 (07:57):
Sure, yeah, no, very impressive, although I would also say
that that feeling regresses to the mean pretty quickly. Right
that if there was a place where there was like
some kind of weird localized gravity storm, and you took
a big pile of leaves and you threw it in
the air and it fell down, in a sentence, we
would all go throw leaves in the air and look
(08:17):
at the sentences they made for a while, and then
we'd be like, they're just not good sentences.
Speaker 1 (08:25):
Going back to the Internet bubble, obviously Internet bubble crash,
but then very shortly afterwards, Facebook and Google and Amazon emerged.
So even if there is like a big froth cycle
around some of the valuations and companies and new billionaires
who are emerging, I mean, if it's a bubble like
(08:45):
the web, it maybe a short term financial bubble that
will burst, but a fundamental platform shift in terms of
how we interact with technology in the world, Like do
you believe that will be true of AI or didn't
the AI in more similar to web three or to
the metaverse?
Speaker 2 (08:58):
Yeah? Sure, I think AI has some similarities to the
Web bubble as compared to other bubbles of the day.
So you know, a bubble that was roughly concurrent with
the Web bubble was the Enron bubble, right, the energy
trading bubble, and that was just accounting fraud, and when
it was over there was nothing but like indictments and
a language corpus because they couldn't be arked to pay
a lawyer to redact their emails before they submitted them
(09:21):
for discovery, whereas the Web left behind like a couple
million humanities undergraduates who'd been inveigled to drop out and
become Pearl and python in HTML jocks, and so you know,
you say that what succeeded the dot com bubble was
Amazon and Google and Facebook and so on, but there
was an intermediate step there, and the intermediate step was
(09:42):
Web too. It was a ton of little amazing startups
that were what you got when skill practitioners were liberated
from the need to indulge the foolish fantasies of their
bosses who were very good at getting capital and very
bad at coming up with products. So that was pretty amazing.
Although we shoul it's say that the Enron bubble, the
(10:02):
Web bubble, and the AI bubble foundationally are not about
what they produce. They're about separating suckers from their money,
right that like the point of a bubble is to
get insiders to cash out and leave ordinary investors holding
the bag. So no bubble is good, but some bubbles
have productive residues. So in that sense, AI, I think
(10:23):
is like the Web in that we're going to have
a lot of GPUs at ten cents on the dollar,
a lot of skilled practitioners looking for work and maybe
hiring each other. And also these open source models that
will continue to function whether or not there's a company
that produce them, because open source software continues to exist
for as long as it exists. But there's a big difference,
actually two. So the first is the vibe. So if
(10:47):
you go back and you read like Harvard Business Review
articles or NBER reports on the workforce in the era
of the web, what you see are all these articles
about like bosses who have to be dragkicking and screaming
into allowing their workers to use a technology that they
see as transformative. And when you look at comparable reports today,
(11:08):
it's full of bosses threatening to fire workers if they
don't use AI. Right. This is now getting to the
material way in which the web is different from AI.
The unit economics of the web were very good. Every
new user of the web made the web more profitable.
Every new use of the web made the web more profitable,
and every generation of the web was more profitable than
the previous generation. That's the opposite of AI. Right, So
(11:29):
AI companies are selling hundred dollars bills for a dollar apiece.
So if you become an AI customer, they start to
lose money. If you continue to be an AI customer,
they lose more money. And the next generation of AI
is going to be selling two one hundred dollars bills
for a dollar apiece, and so it's going to be worse.
And so that makes it a qualitatively and quantitatively different
(11:51):
kind of bubble to the web bubble.
Speaker 1 (11:54):
So what's your role in all of this? And I
think you mentioned the book or in your speech that
you gave Washington that you know, as a science fiction
right to people always want to ask you if your
opinion on AI because I guess so many of the
AI overlords talk about how science fiction governs their thoughts
about what they're building. You know, you've said as well,
(12:14):
science fiction is anti inevitableist literature, as it was a
great phrase. I mean, what is what is your role
in this? What do you what do you want people
to understand to do differently?
Speaker 2 (12:25):
Yeah, I think that the people who run a bubble,
whether or not it has a productive residue, you always
want to style themselves as avatars of the great forces
of history and the iron laws of economics. Right, they're
not imposing a technology on you. Rather, they are just
doing what the circumstances demand. And if it wasn't them,
(12:46):
it'd be someone else, although it has to be them
because they're extraordinary. But but you know, this is the
moment at which we are going to get AI and
we're all going to use AI, and there is no
future without AI, and the only II that we're going
to have in the future is the AI they're making,
and everybody else just shut up. And I call this
a kind of species of vulgar Thatcherism, because you know
(13:08):
Margaret Thatcher, she had this aphorism, there is no alternative,
and this is the mother of all thoughts stopping cliches,
because what she meant by there is no alternative is
don't you dare try and think of an alternative? Right.
Another way of saying there is no alternative is resistance
is futile, but there is no alternative implies Also, if
(13:29):
you're unhappy with this, don't get mad at me. Get
mad at the iron laws of economics and the great
forces of history, because it could be no other way,
and science fiction is entirely concerned with all the other
ways things could be right and in challenging the idea
that everything is inevitable and nothing is contingent.
Speaker 1 (13:52):
So what's the alternative vision you're laying out for what
the future may be? I mean, right now it looks like,
you know, three trillion dollar IPOs, this year's basic exentropic
open AI. You know, new trillionaires and billionaires and multimillionaires,
all of whom will then have the capital. Well, the
(14:15):
share evaluation is to quiem more companies the cash money
to support politicians and packs like. What might disrupt the
futures is currently emerging.
Speaker 2 (14:27):
So I can't predict when the bubble will pop. But
firms that are making tens of billions of dollars a
year and spending hundreds of billions of dollars a year
and that are telling us they need to spend more
hundreds of billions of dollars a year, but whenever they
try to raise prices. I mean, Anthropic just tried to
clean up its balance sheet a little by saying we're
not selling hundred dollar bills for a dollar apiece anymore.
(14:48):
They're going to be five dollars apiece, and all their
best customers are like, I guess we can't use AI
anymore because at five dollars, these hundred dollar bills are
not worth it. So when they run out of peace,
people who will give the money. Because listed are privately held,
you can't spend more than your taking in unless you
have a creditor. Right, it doesn't matter how big your
(15:10):
IPO was, Right, you have operating expenses. I mean, I
guess you could just like loan the company money from
your own shares, but you know this kicks off the
crunch where you flog your shares to get operating capital
for the company. The supply of shares goes up, the
value of the shares goes down, and you're in a
death spiral. So you know, I don't know what's going
(15:32):
to shatter this fragile system we're in. Maybe it's like
Golf States say, sorry, Dario, I know you told us
that we were going to make God here, but we've
decided that we need to have LNG terminals because Donald
Trump keeps getting the Iranians to blow ours up, and
we don't think we have a future without LNG terminals,
(15:53):
even if we are going to own eight percent of God,
and then, like without the golf, you lose the the
rest of the financing collapses. Who knows, right, it could
be any or all of the above, but anything that
can't go on forever eventually stops.
Speaker 1 (16:08):
I think, I think, I guess that the pushback would
be to say, well, maybe right, but you know, if
there's there are other companies like Google, for example, who
can continue to fund their plans which are very similar
to Anthropic and opening eyes into SpaceX is because of
their you know, they're they're they're profitably if they're operating
operating business, and that like somebody may they all Elon
(16:31):
may be able to convince the public and public markets
believing for long enough that he can bring together space
and robotics and data centers and all these types of things.
I mean, I guess like the proof will being the
pulling is it's just unknowable, but certainly that is what
is clear is that the it's a historical disconnect between
the financial performance of these companies and the evaluation and
(16:54):
and and I mean, that's that's unarguable. After the break,
what if a computer program decided whether or not you
(17:14):
deserve a job stay with us. We sort of spent
quite a lot of time I think setting the stakes,
the backdrop. There's two more things I want to talk
about while we're together. One is people and the other
is politics. But we've had too much fun to define
(17:35):
what a reverse centaur is.
Speaker 2 (17:38):
Yeah, well, I think that in a long history of automation,
where automation is driven by workers, it tends to be
about improving quality, whereas capital driven automation is tilted towards
improving throughput. Not because bosses are mustache twirling villains, but
because when you have an asset that's depreciating, you want
to get as much use out of it as you
(17:59):
can before it falls off balance sheet. Another aspect that's
closely related in automation theory is the idea of a centaur,
which is a person assisted by a machine. So that's
you on a bicycle, you with a spell checker, you
with an ide right, Like when I started programming, we
didn't have debuggers, right, So like that makes you a
(18:19):
centaur too, whereas a reverse centaur is a person conscripted
to serve as a peripheral for a machine, and because
of those dynamics of how capital views its investments, that
person is usually the bottleneck in the machine's performance. Right.
People generally can't work as long or as fast as
(18:39):
a machine, and that person is now worked at the
edge of their capability and to the edge of their endurance,
because that's how you maximize the return on the hard
asset that they are assisting. And again this is not
a new idea, Like there's a reason that you know,
whether you're looking at Charlie Chaplin in Modern Times or
Lucille Ball in that episode where she and Ethel are
(19:01):
putting chocolates in the chocolate box on the conveyor belt,
that the motif of automation is the speed up, right,
and it's it's the worker being used up by the machine.
And you know, the warehouses that are most automated in
America are Amazon's warehouses, and they're also the warehouses with
the highest rate of significant injury. Those aren't coincidences. Those
(19:21):
are like those are co determined, right, because if you're
going to put seven figures into warehouse automation, you want
to work the warehouse automation as quickly as you can.
So an Amazon driver is there because even if you
could get the van to drive itself in urban traffic, well,
it couldn't get the parcel onto your porch or onto
your mailbox or you know, into your office building, and
(19:43):
so it needs to have a driver there, and so
you work that driver at the outer limit of their
ability and their endurance. And the reason people pee in
the vans is because you cannot make the quota unless
you operate at a speed that requires that you not
have kidneys.
Speaker 1 (20:00):
I mean you said in the book that you get
penalized for swerving even if you're avoiding Yeah, a danger.
I mean that does that really true?
Speaker 2 (20:07):
Yeah? So Katie Wells just published a good ethnography. She's
done a lot of work on Amazon drivers and on
algorithmic wage discrimination and algorithmic management. She just published some
ethnographies where she worked with another Amazon a former Amazon driver,
to interview a bunch of current and past Amazon drivers,
including one who was fired for swerving when a big
rig crossed the medium because the driver had fallen asleep
(20:29):
and was headed for a head on collision. And they
were fired for making a dangerous maneuver, which is to say,
swerving out of the way of a truck that would
have killed them. And there's lots of petty examples too, though, right,
you can get dinged on your performance if the camera
thinks that your mouth has opened too much, because singing
is considered distracted driving. You know, it's something of a
(20:50):
paradox that these heavily surveilled workers who you would think
those cameras might be used to vindicate them, find that
those cameras are only an ever used to penalize them.
Right that if you've got a million cameras on your truck,
that's when you show your boss, hey, look I got
I swerve because the truck was coming straight at me.
(21:11):
But that's never in the field. In fact, you never
get to talk to the person who fires you, because
the way that it works with the DSPs is that
they don't they're not Amazon, they're they are themselves so
called independent businesses that hire you. And then the app
tells your boss to fire you. And then you can
tell your boss all day long, I swerve to avoid
(21:32):
a truck, But your boss doesn't get to keep you
on the payroll if Amazon says that they're not allowed
to and Amazon makes that determination based on their own
internal logic.
Speaker 1 (21:44):
Unpack this idea of Amazon telling the DSP to fire
the driver, and the driver is say, why don't you
review the well, you know, the video footage and the
DSP saying, well, even if we could do that, it
wouldn't change our decision. I mean, how is this being exactive?
Because the Amazon worker ethnography stuff is haunting, but has
been preceded for sure the AI error? Right, Like, what's
(22:09):
what's being accelerated by AI? In what you're describing here?
Speaker 2 (22:13):
You know you've hit on one of the points that
I wanted to make, which is that we can understand
the problems of AI and also the potential of AI,
both as being in a lineage and not being with
earlier phenomena. I'm firmly the belief that AI is a
normal technology and not extraordinary, neither extraordinarily evil nor extraordinarily amazing.
It's just a new technology, and it has all the
(22:34):
potential and all the problems that technologies have. The bubble
is extraordinary only because of its scale. And so there
has been so much well publicized conduct by firms that
have made workers redundant or limited how workers worked and
insisted that they work alongside of output from bots with
predictably disastrous results, right like, so not like it. It's
(22:59):
the only way you could be surprised is if you
were deliberately not listening to people who are warning you
why you shouldn't do it. And that is a feature
of the bubble, right, That's a feature first of all
of leaders themselves just being gripped by FOMO and being insecure,
but also the fact that investors are quite excited by
news that a firm has added AI to its operations,
(23:23):
and so there's maybe even a rational basis for harming
the firm's outputs in order to produce a narrative that's
friendly to investors. And it's one of those things where
we were told that if only we made top executives
shareholders of the firm, they would be incentivized to do
what's best for the firm. It actually turns out that
what's best for the firm might involve some short term
(23:45):
pain to the share price. But if your net worth
is tied up and shares in the firm you work for,
then you become as short term as the most nimble,
footloose hedge fund guy who has taken a position for
three weeks and is going to be out again and
on the road and never look back.
Speaker 1 (24:02):
You also talk about sort of how the harm of
technology percolates. You said you have to find people without
social power and you grind down the rough edges on
their bodies, which is haunting phrase, But then it goes
to you in the end that metastasizes. And you've also
(24:22):
said for AI to be valuable, it has to replace
high wage workers. Talk about both of those ideas.
Speaker 2 (24:28):
Yeah, so you're describing something I call the shitty technology
adoption curve, which is basically, when you've got it, when
you've got something you want to do that's terrible and
hurts the people who use it. If you make me
the first person who uses it, unlike a mouthy, affluent
white guy who speaks English as a native language and
has a giant platform, and it will be hard for
you to deploy it. But if you start with people
(24:50):
who don't have social capital, who no one listens to,
and who can't say no, you can both normalize the
technology but also sand down the things that people find
most odious about it. And so you know, if you
think about an example, here would be the way that
algorithmic wage discrimination works. So the first really widespread use
(25:12):
of algorithm wage discrimination was at uber, where they were
making inferences about the economic desperation of drivers based on
whether or not they signed on to take rides that
were bad deals, right, whether they poorly compensated, and then
the drivers who took poorly compensated rides were offered progressively
lower compensation. Right, the algorithm is seeking the floor at
(25:33):
which they'll pay. And so you know uber drivers, we
call them unskilled. I don't think that's fair, but they're
certainly not professionally certified workers. But Katie Wells, again the
woman who's done all this work on algorithmic wage discrimination
and algorithm management, has written about how nurses are now
experiencing this. So you know, hospitals in the US preferentially
(25:57):
higher nurses as contractors because that allows them to do
union avoidance. And it used to be that you would
hire a contract nurse through a body shop, through a
staffing agency in town, and there'd be three or four
or maybe one or two. And now there's just four apps,
and they operate nationally, and they all advertise themselves as
an uber for nursing, and at least two of them,
(26:19):
if not all of them, before they offer a shift
to a nurse, look up the nurse's credit history and
they make inferences about the economic desperation about the nurse
based on how much credit card debt they're carrying and
whether it's delinquent, and they offer a lower wage based
on that. So they extract a desperation premium from nurses
before they're offered a shift. And so you can see
(26:40):
how this starts with Uber drivers, who, if they're not
unskilled at least don't require anything more than a car
and a driving license to get started with Uber, and
goes to nurses who are trained for several years and
have to maintain professional licensure in order to do their job.
And you could imagine that that's going to come for
other kinds of workers too. And so, you know, if
(27:01):
you want to know what the future looks like, I'm
a science fiction writer. I don't know that I can
predict it. But if you want to know what it's
leading indicators are, look to the terrible things that are
being done to people with less social capital as you.
Because if they're profitable to do to people with less
social capitalis you, They'll be even more profitable if they're
done to you. And someone is going to get the
idea of trying to do it now, whether or not
(27:21):
they're successful is something that's socially determined, not inevitable, but
you should imagine that it's going to happen, you know.
William Gibson said, the future is here is just not
evenly distributed. If you want to find out where the
worst aspects of the future are pulling up very thick
and on the ground, look at Amazon warehouse workers and
Uber drivers and immigrants and people in ice detention and
(27:42):
kids and so on. That's where it's all headed.
Speaker 1 (27:45):
And Gibson also said, as I learned from reading the
introduction to a book, the street finds its own use
for things. Yeah, I talk about the anti inevitable. Is
this where politics comes in. I might listen to you
and as recline with Tim Woo and getting into you know,
some some fairly detailed policy discussions or policy solutions. But
is policy and regulation the antidote to all of this?
Speaker 2 (28:08):
Well, the regulation sets the contours on which markets operate.
I will confess that I don't think markets are the
only or necessarily the best way to solve our allocation problems.
But you know, the term free market, as first coined
or popularized at least by David Ricardo, did not mean
a market free from regulation. I meant a market free
from rents, right, A meant a market where factors of
(28:30):
production were freely traded and not owned by financiers. And
so if you wanted to make something, you didn't have
to go rent it. You could just acquire it on
the market or buy it, or another firm would sell
it to you that was doing something productive with it.
So it's basically the end of landlords. That's that's what
a free market was, whether that was landlords for ideas
in the form of patents or landlords for property or
(28:50):
what have you. So that's the origin of the idea
of a free market. And so every market, whether it's
the markets that we call heavily regulator to the markets
that we call regulated, is determined by regulation, by what
we allow firms to do, by what we prohibit from firms,
and by the extent to which firm's conduct can be
(29:11):
practically speaking policed. You know, it's very hard for the
referee to ensure a fair game if the players are
more powerful than the ref What do you do with that?
Speaker 1 (29:20):
I guess because like it's tempting to ground these conversations
in like AI and technology and big technology, but they
sort of the set of problems you describe and we're
talking about are problems of how society is organized by
government and what role government plays as power broken between
citizens and organizations.
Speaker 2 (29:42):
Right, it really depends on which aspect of AI we're
talking about, Like if we're talking about the question of
AI and creative labor, I don't think copyright solves this.
We keep expanding copyright and then we discover that in
a market where all creative workers have to sell their
work through a handful of firms, you know, five publishers
and four studios and three labels, two companies that do
the apps, and one company does the ebooks and audiobooks,
(30:03):
it doesn't matter how many rights we have, we just
bargain them away. Right. It's like giving bulliked school kids
extra lunch money. It just doesn't matter how much you
give them. They don't get lunch. But we do have
answers to this right that, we've seen successful gambits by
creative workers to resist using AI to erode their wages
and working conditions. The one group of workers who did
it most successfully was the screenwriters. And the big difference
(30:26):
between the screenwriters and every other kind of worker except
for other Hollywood workers is that they are allowed to
do sectoral bargaining, which was outlawed in the Taft Hartley
Act of nineteen forty seven. Which sectoral bargaining is when
all the workers in a sector, like everyone who works
at a fast food restaurant, is organized under a union,
and every fast food owner is bound by the contract
(30:46):
with those workers. So, you know, if you see creative
workers kind of at a crossroads, and on one fork
they can argue for more copyright, which is the thing
their bosses want and no other work in America gives
a shit about. And on the other fork is sectoral bargaining.
I think their bosses would hate and every other worker
in America would benefit from. It seems to me like
that it's just you could just do basy and reasoning
(31:08):
without knowing a single thing about the specifics of these policies.
You know, broadly speaking, you're on the right side of
history when you're on the same side as all the
workers and the opposite side is your boss if you
care about labor rights. When we're talking about data centers,
I mean, I think it's not hard to imagine a
regulatory regime that would make data centers better, like, we
(31:28):
could have rules about the carbon emissions of them. We
could have democratic determination about their water usage and their
energy usage and the noise that they create. We could
have very very strict limits or possibly a total prohibition
on the use of eminent domain to expropriate farmers and
other people who have large tracks of land to build
data centers. I mean, all of those things are like,
(31:49):
none of them are new, they're just like again back
to AI as a normal technology. This does not require
extraordinary new regulations to stop people from you know, polluting
and just uptive job sites near your home in ways
that make your quality of life worse, and that also
sometimes results in your home being stolen from you through
(32:09):
an eminent domain. And if we're worried about the way
that local councils are being uveiled into doing this, we
could also have rules about like taking bribes, which is
basically how they're getting these regulatory things through. It doesn't
require anything extraordinary to make that happen. In terms of
the decision outcomes right where we're using AI to decide
(32:30):
who gets parole or who gets bond in Gaza, or
which people DOGE is going to fire or which people
ice are going to round up or shoot in the
face or send to a concentration camp in El Salvador.
I mean, we can just say those things should be illegal,
and we can say that if someone needs to be fired,
that there should be a normal procedure for firing them,
and that you should have to show evidence and so on,
(32:50):
you have to fire them for cause, and we could
honor the existing union contracts that federal workers have. Again, like,
it doesn't require anything new. It just requires that we
not treat AI as an extraordinary technology that when it
says fire someone, we get to ignore all the rules
about when you can fire someone.
Speaker 1 (33:07):
I mean, I guess the one politician who has put
forth the kind of national platform of you know, the
relationship between government technology is Bernie Sanders with his AI
sovereign Wealth fund, which interesting he Trump is also a
fan of of, as is Simultman. So there's something something
there which maybe should tasty.
Speaker 2 (33:28):
But I mean, look, if AI were profitable, it would
be a great target for a sovereign wealth fund. I
don't know why we would want the money losing this
thing in the world to be in public ownership unless
we thought it was a public good, in which case
it might make sense. I mean, there's a sense in
which schools don't make money. It would be weird if
schools made money, you know, elementary schools, public schools. But
I don't think that's what Bernie means. I think he means,
(33:49):
I think these are going to throw off extraordinary amounts
of profit. I just think he's wrong. I think he's mistaken.
And that's getting back to being a better AI critic.
I think being a better AI critic means not accepting
the facially the claims that AI is very valuable and
profitable or will be shortly and requiring at least ordinary
evidence for such an extraordinary claim, for a way that
(34:10):
a company that can't raise the price of its one
hundred dollar bills from one dollars to five dollars without
losing most of its customers will someday be profitable.
Speaker 1 (34:19):
You close the book with this line. The future can
be ours if we never stop remembering that the most
important fact about technology isn't what it does, it's who
it does it for and who it does it to.
What do we do with that? What should our listeners?
What should they do this afternoon tomorrow? Having heard this conversation,
(34:39):
you have to join a polity.
Speaker 2 (34:40):
I wish, because hell truly is other people. I wish
we could all be as solive cystic as the AI
bros want us to be, and we could just solve
things with our individual action. We could just go like
shop our way out of a monopoly and recycle our
way out of the wildfires and so on. Alas that
is not possible. Systemic problems require system solutions, and the
(35:01):
way that you make systemic solutions is being part of
a polity. And that would mean joining the Electronic Frontier Foundation,
getting involved in local politics, getting involved with unions and
your job site. And if there isn't one and you
work in tech, look up Tech Solidarity and the Tech
Workers Coalition. You know, that's that's how you make a difference,
right as being part of a group. And I'm not
(35:23):
saying that because I think it's easier fun to work
in groups. It can be, but like it is just
outrageous how hard it is to convince other people that
you are right. You know, I mean, as much as
I insist on it it as an Agelino, people just
will not listen to me when I say when I
need to get on the five. That's when everyone else
should get the hell off the five. But you know
(35:43):
the fact is that anytime you want to do something superhuman,
something that exceeds the capability of one person, you need
to find a way to work with someone else to
do something that two or more people can do. And
the only way to do that is to be part
of a polity ry. Doctor.
Speaker 1 (36:00):
Oh, thank you, thank you for tech Stuff. I'm oz Veloscian.
This episode was produced by Eliza Dennis and Melissa Slaughter.
It was executive produced by me Julian Nutter and Kate
Osborne for Kaleidoscope and Katrin Norvel for iHeart Podcasts. Jack
(36:24):
Insley mixed this episode and Kyle Murdoch wrote out theme song.
Please do rate and review the show where ever you listen,
and reach out to us at tech Stuff podcast at
gmail dot com.
Welcome to tech Stuff. I'm Os Voloshen. If you've been
listening to this podcast, or if you spend any time
on the Internet, you won't be able to miss the
word in shitification. And our guest today is the man
who coined it, Corey doctor Ol. He's a science fiction author,
a technology activist, and a journalist. His new book is
called The Reverse Centaur's Guide to Life After AI, How
(00:39):
to Think about Artificial Intelligence before It's too late. Corey,
Welcome to text Stuff.
Speaker 2 (00:44):
Thanks as it's a pleasure to be on.
Speaker 1 (00:46):
For anyone who did miss it. What is in ghentification?
Speaker 2 (00:49):
Well, I've worked for the Electronic Frontier Foundation, which is
an NGO, for twenty five years on the question digital rights,
and getting people to engage with that question is very
hard because digital rights questions are abstract and technical and
relate to things that are going to happen in the future,
and for very good reasons, people mostly care about concrete
things happening right now. And what I have done over
(01:12):
the years has come up with different framing devices and
similes and metaphors and narratives too. I'm a science fiction novelist,
but it turned out that a dirty word initification was
the way to get people to engage in it, and
so more specifically in sitification is this theory and description.
It describes how the platforms that we rely on have
turned into piles of shit, where first they are good
(01:34):
to their end users but lock them in, and then
having locked in the end users and made it difficult
for them to leave, they make it worse for them
to make it tempting for businesses. And then when the
businesses are also locked in, they also extract everything they
can from those businesses and line their own pockets and
turn into a pile of shit. But the more interesting
part of it is why they're doing it, And this
(01:54):
maybe relates to the conversation we're having today, which is
about the economics and political economy of monopoly, that when
you let firms get too big to fail, they become
too big to jail, and then that makes them too
big to care, and they destroy our lives for the
same reason your dog looks its balls because they can
and no one makes them start.
Speaker 1 (02:13):
I want to get into all of that with you,
but I've read the acknowledgments of your book, not just
the acknowledgements. If I'm always interesting acknowledgments, and you had
the same editor for this book, this new book, the
Reverse Centaur's Guide, as you did for last year's in
Certification book, and you kind of credit him with helping you,
not just with in Certification argument, but with this argument.
(02:34):
And I'm curious, like, where does the reverse central sit
in the lineage of inertifications? What is it a development
of the same argument or is it totally new argument?
Because in Certification was essentially about the tech platforms of
the Internet, and this is about the experience of AI.
Speaker 2 (02:53):
The common lineage between in Certification and my critique of
AI is grounded in the dysfunctions and pathologies of firms
that conquer their markets of saturate their markets. Monopolies, cartels doopolies.
And one of the consequences of having monopoly is that
(03:15):
you become ungovernable in the worst sense, not in the
cool sense of like a raccoon, but in the bad
sense of like Donald Trump. And you are so liberated
from consequence that you can do unlimited bad things to
lots of people, including people you would think would have
some power in the economy, and just get away with it.
(03:36):
And really the worst ideas of the worst people become
the most profitable way of doing business. So that's like
in shitification at its theoretical core. But reverse centaur is
about another epiphenomenon, another thing that a consequence of monopoly,
which is that if your firm that has saturated its
(03:57):
market right Google has a ninety percent market share, then
you can't grow anymore. And you know, there's this saying
among leftists and crunchy granola types that endless growth is
the ideology of a tumor, and that grounds it in
this idea that the reason firms want to continue growing
(04:18):
as ideological and not practical or concrete. And while there
is some ideology at work there in the way that
we think about our economy and in the environmental consequences
of that, there are very practical, material reasons that companies
want to be perceived as growing and not as mature
and at the end of their growth cycle. And that
(04:40):
is because the share price of a company that is
growing is much higher based on the total economic turnover
of that company than the share price of affirm with
the same turnover. But that is static, and that's not irrational.
A share and affirm is a claim on its future earnings.
Firms that are growing have more future earnings to return
(05:03):
to you than firms that are not growing. And the
corollary of this is that when your firm stops growing,
it becomes overvalued because the future earnings have contracted when
you reach the top of your game and you stop growing,
and that provokes these mass sell offs by investors.
Speaker 1 (05:25):
You mentioned, I think you mentioned that you know the
Invidia cell of when the deep Seak moment happened. Obviously, Yeah,
in video rebounded from that, but it raised the it
raised a specter that growth may not be unlimited.
Speaker 2 (05:37):
Yeah, the video sell off was the largest decapitalization in
twenty four hours of any firm. And when your firm
becomes illiquid, right when it's not, when it's not traded
at this high multiple and so liquid, you can't grow
by buying other firms by offering them stock. You have
to give them money. And money is an exogenous substance
that's produced outside the firm, and shares are things that
(05:58):
are produced inside the firm by typing zeros and to
a spreadsheet. And so it's always preferable to be able
to acquire firms and talent using shares. This endogenous substance
you can produce on demand, and you can only do
that while you're growing. So this is another kind of
growth paradox, right, which is that it's easy to keep
growing when you're growing because you can buy other companies,
(06:19):
but it's hard when you stop growing to start growing again,
not least because all those key employees that have been
paid in shares suddenly see their net worthfall off a
cliff when you when your share price declines, and they
might go look for work elsewhere, which means they're not
going to help you start the firm up again. They're
not going to help you get a back on a
growth trajectory.
Speaker 1 (06:37):
So that the central argument of the book is that
a lot of technology companies order the practice order. The
technology companies have an extraordinary incentive to persuade the white
world that in a AI is inevitable, and that that
kind of mythology is of the essence to protecting their
value and therefore keeping their employees and their ability to
(07:00):
keep growing by buying other companies with their stock.
Speaker 2 (07:04):
Yes, So what I would say is that the book
is exploring why tech bubbles exist at all. And this
is not the first one we've had, you know, metaverse
and cryptocurrency and web three, all those other things.
Speaker 1 (07:16):
But they were all palpably confections, right, those those were
not things that people use, whereas, like all you do
is any conversation you over hear in the subway, in
a restaurant, whatever, is people telling each other how they
use AI. So that arguably is a big difference.
Speaker 2 (07:31):
So it's just so those bubbles were smaller. Why is
the a bubble larger? And some of it is because
AI is realer than those things. So I'm a fake
computer scientist. I have an honorary doctor in computer science
from the Open University, and in my capacity as a
fake computer scientist, I can tell you that AI is
pretty interesting.
Speaker 1 (07:50):
And from a user point of view, I would argue
a feeling of technology magic for the first time in
a long time.
Speaker 2 (07:57):
Sure, yeah, no, very impressive, although I would also say
that that feeling regresses to the mean pretty quickly. Right
that if there was a place where there was like
some kind of weird localized gravity storm, and you took
a big pile of leaves and you threw it in
the air and it fell down, in a sentence, we
would all go throw leaves in the air and look
(08:17):
at the sentences they made for a while, and then
we'd be like, they're just not good sentences.
Speaker 1 (08:25):
Going back to the Internet bubble, obviously Internet bubble crash,
but then very shortly afterwards, Facebook and Google and Amazon emerged.
So even if there is like a big froth cycle
around some of the valuations and companies and new billionaires
who are emerging, I mean, if it's a bubble like
(08:45):
the web, it maybe a short term financial bubble that
will burst, but a fundamental platform shift in terms of
how we interact with technology in the world, Like do
you believe that will be true of AI or didn't
the AI in more similar to web three or to
the metaverse?
Speaker 2 (08:58):
Yeah? Sure, I think AI has some similarities to the
Web bubble as compared to other bubbles of the day.
So you know, a bubble that was roughly concurrent with
the Web bubble was the Enron bubble, right, the energy
trading bubble, and that was just accounting fraud, and when
it was over there was nothing but like indictments and
a language corpus because they couldn't be arked to pay
a lawyer to redact their emails before they submitted them
(09:21):
for discovery, whereas the Web left behind like a couple
million humanities undergraduates who'd been inveigled to drop out and
become Pearl and python in HTML jocks, and so you know,
you say that what succeeded the dot com bubble was
Amazon and Google and Facebook and so on, but there
was an intermediate step there, and the intermediate step was
(09:42):
Web too. It was a ton of little amazing startups
that were what you got when skill practitioners were liberated
from the need to indulge the foolish fantasies of their
bosses who were very good at getting capital and very
bad at coming up with products. So that was pretty amazing.
Although we shoul it's say that the Enron bubble, the
(10:02):
Web bubble, and the AI bubble foundationally are not about
what they produce. They're about separating suckers from their money,
right that like the point of a bubble is to
get insiders to cash out and leave ordinary investors holding
the bag. So no bubble is good, but some bubbles
have productive residues. So in that sense, AI, I think
(10:23):
is like the Web in that we're going to have
a lot of GPUs at ten cents on the dollar,
a lot of skilled practitioners looking for work and maybe
hiring each other. And also these open source models that
will continue to function whether or not there's a company
that produce them, because open source software continues to exist
for as long as it exists. But there's a big difference,
actually two. So the first is the vibe. So if
(10:47):
you go back and you read like Harvard Business Review
articles or NBER reports on the workforce in the era
of the web, what you see are all these articles
about like bosses who have to be dragkicking and screaming
into allowing their workers to use a technology that they
see as transformative. And when you look at comparable reports today,
(11:08):
it's full of bosses threatening to fire workers if they
don't use AI. Right. This is now getting to the
material way in which the web is different from AI.
The unit economics of the web were very good. Every
new user of the web made the web more profitable.
Every new use of the web made the web more profitable,
and every generation of the web was more profitable than
the previous generation. That's the opposite of AI. Right, So
(11:29):
AI companies are selling hundred dollars bills for a dollar apiece.
So if you become an AI customer, they start to
lose money. If you continue to be an AI customer,
they lose more money. And the next generation of AI
is going to be selling two one hundred dollars bills
for a dollar apiece, and so it's going to be worse.
And so that makes it a qualitatively and quantitatively different
(11:51):
kind of bubble to the web bubble.
Speaker 1 (11:54):
So what's your role in all of this? And I
think you mentioned the book or in your speech that
you gave Washington that you know, as a science fiction
right to people always want to ask you if your
opinion on AI because I guess so many of the
AI overlords talk about how science fiction governs their thoughts
about what they're building. You know, you've said as well,
(12:14):
science fiction is anti inevitableist literature, as it was a
great phrase. I mean, what is what is your role
in this? What do you what do you want people
to understand to do differently?
Speaker 2 (12:25):
Yeah, I think that the people who run a bubble,
whether or not it has a productive residue, you always
want to style themselves as avatars of the great forces
of history and the iron laws of economics. Right, they're
not imposing a technology on you. Rather, they are just
doing what the circumstances demand. And if it wasn't them,
(12:46):
it'd be someone else, although it has to be them
because they're extraordinary. But but you know, this is the
moment at which we are going to get AI and
we're all going to use AI, and there is no
future without AI, and the only II that we're going
to have in the future is the AI they're making,
and everybody else just shut up. And I call this
a kind of species of vulgar Thatcherism, because you know
(13:08):
Margaret Thatcher, she had this aphorism, there is no alternative,
and this is the mother of all thoughts stopping cliches,
because what she meant by there is no alternative is
don't you dare try and think of an alternative? Right.
Another way of saying there is no alternative is resistance
is futile, but there is no alternative implies Also, if
(13:29):
you're unhappy with this, don't get mad at me. Get
mad at the iron laws of economics and the great
forces of history, because it could be no other way,
and science fiction is entirely concerned with all the other
ways things could be right and in challenging the idea
that everything is inevitable and nothing is contingent.
Speaker 1 (13:52):
So what's the alternative vision you're laying out for what
the future may be? I mean, right now it looks like,
you know, three trillion dollar IPOs, this year's basic exentropic
open AI. You know, new trillionaires and billionaires and multimillionaires,
all of whom will then have the capital. Well, the
(14:15):
share evaluation is to quiem more companies the cash money
to support politicians and packs like. What might disrupt the
futures is currently emerging.
Speaker 2 (14:27):
So I can't predict when the bubble will pop. But
firms that are making tens of billions of dollars a
year and spending hundreds of billions of dollars a year
and that are telling us they need to spend more
hundreds of billions of dollars a year, but whenever they
try to raise prices. I mean, Anthropic just tried to
clean up its balance sheet a little by saying we're
not selling hundred dollar bills for a dollar apiece anymore.
(14:48):
They're going to be five dollars apiece, and all their
best customers are like, I guess we can't use AI
anymore because at five dollars, these hundred dollar bills are
not worth it. So when they run out of peace,
people who will give the money. Because listed are privately held,
you can't spend more than your taking in unless you
have a creditor. Right, it doesn't matter how big your
(15:10):
IPO was, Right, you have operating expenses. I mean, I
guess you could just like loan the company money from
your own shares, but you know this kicks off the
crunch where you flog your shares to get operating capital
for the company. The supply of shares goes up, the
value of the shares goes down, and you're in a
death spiral. So you know, I don't know what's going
(15:32):
to shatter this fragile system we're in. Maybe it's like
Golf States say, sorry, Dario, I know you told us
that we were going to make God here, but we've
decided that we need to have LNG terminals because Donald
Trump keeps getting the Iranians to blow ours up, and
we don't think we have a future without LNG terminals,
(15:53):
even if we are going to own eight percent of God,
and then, like without the golf, you lose the the
rest of the financing collapses. Who knows, right, it could
be any or all of the above, but anything that
can't go on forever eventually stops.
Speaker 1 (16:08):
I think, I think, I guess that the pushback would
be to say, well, maybe right, but you know, if
there's there are other companies like Google, for example, who
can continue to fund their plans which are very similar
to Anthropic and opening eyes into SpaceX is because of
their you know, they're they're they're profitably if they're operating
operating business, and that like somebody may they all Elon
(16:31):
may be able to convince the public and public markets
believing for long enough that he can bring together space
and robotics and data centers and all these types of things.
I mean, I guess like the proof will being the
pulling is it's just unknowable, but certainly that is what
is clear is that the it's a historical disconnect between
the financial performance of these companies and the evaluation and
(16:54):
and and I mean, that's that's unarguable. After the break,
what if a computer program decided whether or not you
(17:14):
deserve a job stay with us. We sort of spent
quite a lot of time I think setting the stakes,
the backdrop. There's two more things I want to talk
about while we're together. One is people and the other
is politics. But we've had too much fun to define
(17:35):
what a reverse centaur is.
Speaker 2 (17:38):
Yeah, well, I think that in a long history of automation,
where automation is driven by workers, it tends to be
about improving quality, whereas capital driven automation is tilted towards
improving throughput. Not because bosses are mustache twirling villains, but
because when you have an asset that's depreciating, you want
to get as much use out of it as you
(17:59):
can before it falls off balance sheet. Another aspect that's
closely related in automation theory is the idea of a centaur,
which is a person assisted by a machine. So that's
you on a bicycle, you with a spell checker, you
with an ide right, Like when I started programming, we
didn't have debuggers, right, So like that makes you a
(18:19):
centaur too, whereas a reverse centaur is a person conscripted
to serve as a peripheral for a machine, and because
of those dynamics of how capital views its investments, that
person is usually the bottleneck in the machine's performance. Right.
People generally can't work as long or as fast as
(18:39):
a machine, and that person is now worked at the
edge of their capability and to the edge of their endurance,
because that's how you maximize the return on the hard
asset that they are assisting. And again this is not
a new idea, Like there's a reason that you know,
whether you're looking at Charlie Chaplin in Modern Times or
Lucille Ball in that episode where she and Ethel are
(19:01):
putting chocolates in the chocolate box on the conveyor belt,
that the motif of automation is the speed up, right,
and it's it's the worker being used up by the machine.
And you know, the warehouses that are most automated in
America are Amazon's warehouses, and they're also the warehouses with
the highest rate of significant injury. Those aren't coincidences. Those
(19:21):
are like those are co determined, right, because if you're
going to put seven figures into warehouse automation, you want
to work the warehouse automation as quickly as you can.
So an Amazon driver is there because even if you
could get the van to drive itself in urban traffic, well,
it couldn't get the parcel onto your porch or onto
your mailbox or you know, into your office building, and
(19:43):
so it needs to have a driver there, and so
you work that driver at the outer limit of their
ability and their endurance. And the reason people pee in
the vans is because you cannot make the quota unless
you operate at a speed that requires that you not
have kidneys.
Speaker 1 (20:00):
I mean you said in the book that you get
penalized for swerving even if you're avoiding Yeah, a danger.
I mean that does that really true?
Speaker 2 (20:07):
Yeah? So Katie Wells just published a good ethnography. She's
done a lot of work on Amazon drivers and on
algorithmic wage discrimination and algorithmic management. She just published some
ethnographies where she worked with another Amazon a former Amazon driver,
to interview a bunch of current and past Amazon drivers,
including one who was fired for swerving when a big
rig crossed the medium because the driver had fallen asleep
(20:29):
and was headed for a head on collision. And they
were fired for making a dangerous maneuver, which is to say,
swerving out of the way of a truck that would
have killed them. And there's lots of petty examples too, though, right,
you can get dinged on your performance if the camera
thinks that your mouth has opened too much, because singing
is considered distracted driving. You know, it's something of a
(20:50):
paradox that these heavily surveilled workers who you would think
those cameras might be used to vindicate them, find that
those cameras are only an ever used to penalize them.
Right that if you've got a million cameras on your truck,
that's when you show your boss, hey, look I got
I swerve because the truck was coming straight at me.
(21:11):
But that's never in the field. In fact, you never
get to talk to the person who fires you, because
the way that it works with the DSPs is that
they don't they're not Amazon, they're they are themselves so
called independent businesses that hire you. And then the app
tells your boss to fire you. And then you can
tell your boss all day long, I swerve to avoid
(21:32):
a truck, But your boss doesn't get to keep you
on the payroll if Amazon says that they're not allowed
to and Amazon makes that determination based on their own
internal logic.
Speaker 1 (21:44):
Unpack this idea of Amazon telling the DSP to fire
the driver, and the driver is say, why don't you
review the well, you know, the video footage and the
DSP saying, well, even if we could do that, it
wouldn't change our decision. I mean, how is this being exactive?
Because the Amazon worker ethnography stuff is haunting, but has
been preceded for sure the AI error? Right, Like, what's
(22:09):
what's being accelerated by AI? In what you're describing here?
Speaker 2 (22:13):
You know you've hit on one of the points that
I wanted to make, which is that we can understand
the problems of AI and also the potential of AI,
both as being in a lineage and not being with
earlier phenomena. I'm firmly the belief that AI is a
normal technology and not extraordinary, neither extraordinarily evil nor extraordinarily amazing.
It's just a new technology, and it has all the
(22:34):
potential and all the problems that technologies have. The bubble
is extraordinary only because of its scale. And so there
has been so much well publicized conduct by firms that
have made workers redundant or limited how workers worked and
insisted that they work alongside of output from bots with
predictably disastrous results, right like, so not like it. It's
(22:59):
the only way you could be surprised is if you
were deliberately not listening to people who are warning you
why you shouldn't do it. And that is a feature
of the bubble, right, That's a feature first of all
of leaders themselves just being gripped by FOMO and being insecure,
but also the fact that investors are quite excited by
news that a firm has added AI to its operations,
(23:23):
and so there's maybe even a rational basis for harming
the firm's outputs in order to produce a narrative that's
friendly to investors. And it's one of those things where
we were told that if only we made top executives
shareholders of the firm, they would be incentivized to do
what's best for the firm. It actually turns out that
what's best for the firm might involve some short term
(23:45):
pain to the share price. But if your net worth
is tied up and shares in the firm you work for,
then you become as short term as the most nimble,
footloose hedge fund guy who has taken a position for
three weeks and is going to be out again and
on the road and never look back.
Speaker 1 (24:02):
You also talk about sort of how the harm of
technology percolates. You said you have to find people without
social power and you grind down the rough edges on
their bodies, which is haunting phrase, But then it goes
to you in the end that metastasizes. And you've also
(24:22):
said for AI to be valuable, it has to replace
high wage workers. Talk about both of those ideas.
Speaker 2 (24:28):
Yeah, so you're describing something I call the shitty technology
adoption curve, which is basically, when you've got it, when
you've got something you want to do that's terrible and
hurts the people who use it. If you make me
the first person who uses it, unlike a mouthy, affluent
white guy who speaks English as a native language and
has a giant platform, and it will be hard for
you to deploy it. But if you start with people
(24:50):
who don't have social capital, who no one listens to,
and who can't say no, you can both normalize the
technology but also sand down the things that people find
most odious about it. And so you know, if you
think about an example, here would be the way that
algorithmic wage discrimination works. So the first really widespread use
(25:12):
of algorithm wage discrimination was at uber, where they were
making inferences about the economic desperation of drivers based on
whether or not they signed on to take rides that
were bad deals, right, whether they poorly compensated, and then
the drivers who took poorly compensated rides were offered progressively
lower compensation. Right, the algorithm is seeking the floor at
(25:33):
which they'll pay. And so you know uber drivers, we
call them unskilled. I don't think that's fair, but they're
certainly not professionally certified workers. But Katie Wells, again the
woman who's done all this work on algorithmic wage discrimination
and algorithm management, has written about how nurses are now
experiencing this. So you know, hospitals in the US preferentially
(25:57):
higher nurses as contractors because that allows them to do
union avoidance. And it used to be that you would
hire a contract nurse through a body shop, through a
staffing agency in town, and there'd be three or four
or maybe one or two. And now there's just four apps,
and they operate nationally, and they all advertise themselves as
an uber for nursing, and at least two of them,
(26:19):
if not all of them, before they offer a shift
to a nurse, look up the nurse's credit history and
they make inferences about the economic desperation about the nurse
based on how much credit card debt they're carrying and
whether it's delinquent, and they offer a lower wage based
on that. So they extract a desperation premium from nurses
before they're offered a shift. And so you can see
(26:40):
how this starts with Uber drivers, who, if they're not
unskilled at least don't require anything more than a car
and a driving license to get started with Uber, and
goes to nurses who are trained for several years and
have to maintain professional licensure in order to do their job.
And you could imagine that that's going to come for
other kinds of workers too. And so, you know, if
(27:01):
you want to know what the future looks like, I'm
a science fiction writer. I don't know that I can
predict it. But if you want to know what it's
leading indicators are, look to the terrible things that are
being done to people with less social capital as you.
Because if they're profitable to do to people with less
social capitalis you, They'll be even more profitable if they're
done to you. And someone is going to get the
idea of trying to do it now, whether or not
(27:21):
they're successful is something that's socially determined, not inevitable, but
you should imagine that it's going to happen, you know.
William Gibson said, the future is here is just not
evenly distributed. If you want to find out where the
worst aspects of the future are pulling up very thick
and on the ground, look at Amazon warehouse workers and
Uber drivers and immigrants and people in ice detention and
(27:42):
kids and so on. That's where it's all headed.
Speaker 1 (27:45):
And Gibson also said, as I learned from reading the
introduction to a book, the street finds its own use
for things. Yeah, I talk about the anti inevitable. Is
this where politics comes in. I might listen to you
and as recline with Tim Woo and getting into you know,
some some fairly detailed policy discussions or policy solutions. But
is policy and regulation the antidote to all of this?
Speaker 2 (28:08):
Well, the regulation sets the contours on which markets operate.
I will confess that I don't think markets are the
only or necessarily the best way to solve our allocation problems.
But you know, the term free market, as first coined
or popularized at least by David Ricardo, did not mean
a market free from regulation. I meant a market free
from rents, right, A meant a market where factors of
(28:30):
production were freely traded and not owned by financiers. And
so if you wanted to make something, you didn't have
to go rent it. You could just acquire it on
the market or buy it, or another firm would sell
it to you that was doing something productive with it.
So it's basically the end of landlords. That's that's what
a free market was, whether that was landlords for ideas
in the form of patents or landlords for property or
(28:50):
what have you. So that's the origin of the idea
of a free market. And so every market, whether it's
the markets that we call heavily regulator to the markets
that we call regulated, is determined by regulation, by what
we allow firms to do, by what we prohibit from firms,
and by the extent to which firm's conduct can be
(29:11):
practically speaking policed. You know, it's very hard for the
referee to ensure a fair game if the players are
more powerful than the ref What do you do with that?
Speaker 1 (29:20):
I guess because like it's tempting to ground these conversations
in like AI and technology and big technology, but they
sort of the set of problems you describe and we're
talking about are problems of how society is organized by
government and what role government plays as power broken between
citizens and organizations.
Speaker 2 (29:42):
Right, it really depends on which aspect of AI we're
talking about, Like if we're talking about the question of
AI and creative labor, I don't think copyright solves this.
We keep expanding copyright and then we discover that in
a market where all creative workers have to sell their
work through a handful of firms, you know, five publishers
and four studios and three labels, two companies that do
the apps, and one company does the ebooks and audiobooks,
(30:03):
it doesn't matter how many rights we have, we just
bargain them away. Right. It's like giving bulliked school kids
extra lunch money. It just doesn't matter how much you
give them. They don't get lunch. But we do have
answers to this right that, we've seen successful gambits by
creative workers to resist using AI to erode their wages
and working conditions. The one group of workers who did
it most successfully was the screenwriters. And the big difference
(30:26):
between the screenwriters and every other kind of worker except
for other Hollywood workers is that they are allowed to
do sectoral bargaining, which was outlawed in the Taft Hartley
Act of nineteen forty seven. Which sectoral bargaining is when
all the workers in a sector, like everyone who works
at a fast food restaurant, is organized under a union,
and every fast food owner is bound by the contract
(30:46):
with those workers. So, you know, if you see creative
workers kind of at a crossroads, and on one fork
they can argue for more copyright, which is the thing
their bosses want and no other work in America gives
a shit about. And on the other fork is sectoral bargaining.
I think their bosses would hate and every other worker
in America would benefit from. It seems to me like
that it's just you could just do basy and reasoning
(31:08):
without knowing a single thing about the specifics of these policies.
You know, broadly speaking, you're on the right side of
history when you're on the same side as all the
workers and the opposite side is your boss if you
care about labor rights. When we're talking about data centers,
I mean, I think it's not hard to imagine a
regulatory regime that would make data centers better, like, we
(31:28):
could have rules about the carbon emissions of them. We
could have democratic determination about their water usage and their
energy usage and the noise that they create. We could
have very very strict limits or possibly a total prohibition
on the use of eminent domain to expropriate farmers and
other people who have large tracks of land to build
data centers. I mean, all of those things are like,
(31:49):
none of them are new, they're just like again back
to AI as a normal technology. This does not require
extraordinary new regulations to stop people from you know, polluting
and just uptive job sites near your home in ways
that make your quality of life worse, and that also
sometimes results in your home being stolen from you through
(32:09):
an eminent domain. And if we're worried about the way
that local councils are being uveiled into doing this, we
could also have rules about like taking bribes, which is
basically how they're getting these regulatory things through. It doesn't
require anything extraordinary to make that happen. In terms of
the decision outcomes right where we're using AI to decide
(32:30):
who gets parole or who gets bond in Gaza, or
which people DOGE is going to fire or which people
ice are going to round up or shoot in the
face or send to a concentration camp in El Salvador.
I mean, we can just say those things should be illegal,
and we can say that if someone needs to be fired,
that there should be a normal procedure for firing them,
and that you should have to show evidence and so on,
(32:50):
you have to fire them for cause, and we could
honor the existing union contracts that federal workers have. Again, like,
it doesn't require anything new. It just requires that we
not treat AI as an extraordinary technology that when it
says fire someone, we get to ignore all the rules
about when you can fire someone.
Speaker 1 (33:07):
I mean, I guess the one politician who has put
forth the kind of national platform of you know, the
relationship between government technology is Bernie Sanders with his AI
sovereign Wealth fund, which interesting he Trump is also a
fan of of, as is Simultman. So there's something something
there which maybe should tasty.
Speaker 2 (33:28):
But I mean, look, if AI were profitable, it would
be a great target for a sovereign wealth fund. I
don't know why we would want the money losing this
thing in the world to be in public ownership unless
we thought it was a public good, in which case
it might make sense. I mean, there's a sense in
which schools don't make money. It would be weird if
schools made money, you know, elementary schools, public schools. But
I don't think that's what Bernie means. I think he means,
(33:49):
I think these are going to throw off extraordinary amounts
of profit. I just think he's wrong. I think he's mistaken.
And that's getting back to being a better AI critic.
I think being a better AI critic means not accepting
the facially the claims that AI is very valuable and
profitable or will be shortly and requiring at least ordinary
evidence for such an extraordinary claim, for a way that
(34:10):
a company that can't raise the price of its one
hundred dollar bills from one dollars to five dollars without
losing most of its customers will someday be profitable.
Speaker 1 (34:19):
You close the book with this line. The future can
be ours if we never stop remembering that the most
important fact about technology isn't what it does, it's who
it does it for and who it does it to.
What do we do with that? What should our listeners?
What should they do this afternoon tomorrow? Having heard this conversation,
(34:39):
you have to join a polity.
Speaker 2 (34:40):
I wish, because hell truly is other people. I wish
we could all be as solive cystic as the AI
bros want us to be, and we could just solve
things with our individual action. We could just go like
shop our way out of a monopoly and recycle our
way out of the wildfires and so on. Alas that
is not possible. Systemic problems require system solutions, and the
(35:01):
way that you make systemic solutions is being part of
a polity. And that would mean joining the Electronic Frontier Foundation,
getting involved in local politics, getting involved with unions and
your job site. And if there isn't one and you
work in tech, look up Tech Solidarity and the Tech
Workers Coalition. You know, that's that's how you make a difference,
right as being part of a group. And I'm not
(35:23):
saying that because I think it's easier fun to work
in groups. It can be, but like it is just
outrageous how hard it is to convince other people that
you are right. You know, I mean, as much as
I insist on it it as an Agelino, people just
will not listen to me when I say when I
need to get on the five. That's when everyone else
should get the hell off the five. But you know
(35:43):
the fact is that anytime you want to do something superhuman,
something that exceeds the capability of one person, you need
to find a way to work with someone else to
do something that two or more people can do. And
the only way to do that is to be part
of a polity ry. Doctor.
Speaker 1 (36:00):
Oh, thank you, thank you for tech Stuff. I'm oz Veloscian.
This episode was produced by Eliza Dennis and Melissa Slaughter.
It was executive produced by me Julian Nutter and Kate
Osborne for Kaleidoscope and Katrin Norvel for iHeart Podcasts. Jack
(36:24):
Insley mixed this episode and Kyle Murdoch wrote out theme song.
Please do rate and review the show where ever you listen,
and reach out to us at tech Stuff podcast at
gmail dot com.
TechStuff News
Hosts And Creators
Oz Woloshyn
Karah Preiss
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Betrayal Weekly
Betrayal Weekly is back for a new season. Every Thursday, Betrayal Weekly shares first-hand accounts of broken trust, shocking deceptions, and the trail of destruction they leave behind. Hosted by Andrea Gunning, this weekly ongoing series digs into real-life stories of betrayal and the aftermath. From stories of double lives to dark discoveries, these are cautionary tales and accounts of resilience against all odds. From the producers of the critically acclaimed Betrayal series, Betrayal Weekly drops new episodes every Thursday. If you would like to share your story, you can reach out to the Betrayal Team by emailing them at betrayalpod@gmail.com and follow us on Instagram at @betrayalpod and @glasspodcasts. Please join our Substack for additional exclusive content, curated book recommendations, and community discussions. Sign up FREE by clicking this link Beyond Betrayal Substack. Join our community dedicated to truth, resilience, and healing. Your voice matters! Be a part of our Betrayal journey on Substack.