Is Artificial Intelligence the Next
Great Bubble?
26.10.202521:00
By
Konstantinos Tsalakos
Author and
digital rights advocate Cory Doctorow speaks to To Vima on the coming AI crash,
the “enshittification” of the Internet, and how Big Tech turned innovation into
decay
What if artificial intelligence isn’t
the most transformative technology in the history of humanity? What if what we
are actually facing is a financial bubble comparable to that of 2008? What if
the services on offer from the major technology platforms don’t improve over
time, but actually get worse—on purpose?
Cory Doctorow is one of the heretics who have no qualms
about answering these questions straight from the hip. As a writer, journalist
and digital rights advocate, he is one of the most dynamic and inquisitive
figures who engage systematically with technology issues.
To Vima spoke with him about the
promises technology has failed to keep, and how we have been led inexorably
down the road to what he has provocatively dubbed the “enshittification” of the
Internet.
Is artificial intelligence the new financial bubble? Is our
reality going to be like something from the Book of Revelations before we know
it?
“Absolutely. The question is whether we can avoid at least
some of the fallout and stop throwing good money after bad, given that every
dollar invested in AI is a dollar wasted. We’re talking about a sector that has
had 700 billion dollars of capital expenditure sunk in it, when by the Tech
Bros’ own pumped-up calculations, it has only generated 60. Which depreciates
its assets in five years, when its executives admit in private that those
assets are worth precisely nothing after three. A sector which, according to
Bain and Company, has to make three trillion dollars by 2030 or face collapse,
and which only ever spends other people’s money. Seven giant AI companies now
account for between 29% and 31% of the US stock market, the S&P 500, and
all of them are losing money. I think the crisis, when it comes, will be worse
than 2008.”00:01
So, what is this: a case of mass hallucination?
“But isn’t that what happens with every bubble? The idea
prevails that there’s no way it could be a bubble, because everyone else is
spending so much money. So I’d better start spending too—either that, or miss
the boat. Bubbles are a way of siphoning off the savings of ordinary savers. I
think a lot of the big AI goliaths may make billions out of this, just like the
titans did in the housing crisis. But most people will lose everything when
artificial intelligence collapses.”
Every bubble is based on a promise. What is the promise
here?
“That artificial intelligence will replace human labor. If
a saleswoman contacts an employer and proposes a product that will allow him to
lay off workers, she’ll definitely have his attention. Bosses are always
enthusiastic about the idea of downsizing. The problem is, there are very few
jobs AI can actually replace, and most of those don’t come with high wage
costs. So even if the salaries of the people it puts out of work were funneled
into the AI company that provided the product that made them redundant, it
still wouldn’t make up for the billions it had spent to develop it.”
An example?
“You can’t really replace programmers with AI, because AI
can write code, but it can’t do software engineering. And, instead of trying to
use AI to provide experienced programmers with the tools they need to become
more productive, the companies are trying to fire all their experienced—and
therefore expensive—programmers and replace them with easily intimidated
younger coders. But, needless to say, the relative beginners aren’t able to
detect the subtle errors artificial intelligence makes (see ‘slopsquatting’).
So, we’re going to see a lot of programmers getting fired. Not because AI can
actually do their jobs, but rather because their boss has convinced themselves
it can.”
So what can we do?
“The best thing we could do would be to travel back in time
and stop funding Open AI. But we can’t. The second-best option is to bring this
gross misallocation of capital to a stop before it’s too late. We’re going to
lose a lot of people who do jobs that matter, and which we rely on, as they’re
replaced by artificial intelligence. And then, when the collapse comes and they
turn AI off, we won’t have anybody to do those jobs. So things are going to be
very bad.”
Can we talk a little bit about your concept of
“enshittification”?
“It’s based on an observation of how digital platforms are
changing for the worse, and offers a theory as to why this is happening.
Platforms are good for the end user, at first. But then they find a way to lock
their end users in, and once they realize that it’s difficult for them to
leave, they downgrade the services they provide them in order to attract more
business customers. And once the business users are locked in, too, they
degrade their offering to them, as well. Finally, in the third stage of the
process, they harvest all the value for themselves, meaning for their
executives and shareholders. These are the three stages of what I call the
‘enshittification” process.
Could you give us an example here, too.
“Think of Google. The image we have is of a tremendously
innovative business. But take a closer look, and you realize that its only
genuinely successful and innovative product is the search engine, which it
created many years ago. All its other successful products were bought, with
each new acquisition raising competition concerns. It took the other kids’ toys
and is playing with them now. The result: Google has captured the search engine
market with a 90% market share. And it used the profits it earned from
acquiring all the companies that could have replaced it one day to hoover up
everything else on the shelves: every operating system, every mobile device,
every provider. Which is how, in 2020, it reached the point where it couldn’t
grow any more, as it already had a 90% market share.”
And so?
“And so, when the US Department of Justice took Google to
court for antitrust practices, the company’s internal memos were released
relating to how the company responded to the growth crisis, when it erupted in
2020. Well, we now know there were two views. On the one hand, there were those
whose focus was on revenues. And that the head of the relevant department,
Prabhakar Raghavan, suggested solving the problem by making Google search
worse. Because making users search more than once to find their answer was more
profitable, since Google would get to show you ads every time you searched. But
while the head of Google search technology, Ben Gomes, hated the idea, his only
counterargument was that he’d feel bad if they went ahead and did it. At the
end of the day, of course, “when money talks, bullshit walks”, as they say.”
Why is this happening?
“It’s due to what I call an ‘enshittified policy
environment’ that allows companies that are bad for consumers to prevail over
companies that provide good services. And it starts with the failure to enforce
antitrust competition law. This leads to regulatory capture, because when a few
companies control most of the market in a sector, they can gang up against the
regulators. In the case of the tech giants, all the other safety valves that
could have prevented the enshittification of their products were also
dismantled—due, once again, to the same over-concentration.”