This is the script of my national network radio report yesterday on
the surprise 500 million dollar AI one month usage bill a major firm
received, and what this tells us about the state of AI hype and slop.
As usual there may have been minor wording variations from this script
as I presented this report live on air.
- - -
So this is the kind of story that the billionaire big tech CEOs would
really probably prefer the public not be thinking about -- so let's
consider it in some depth.
It's certainly annoying if for example you underestimate the cost of
something and when you see the final bill it's a bit higher than you
anticipated. But yeah what about an unexpected bill of around 500
million dollars -- half a BILLION dollars -- over the course of ONE
MONTH. And that's indeed what's been reported to have happened to a
major firm recently with their spending on Anthropic's Claude AI
services. The firm hasn't been publicly identified but at these dollar
amounts it's a fairly constrained set of very large firms that are the
likely candidates.
So what happened and what does this tell us about the ever growing
cesspool of AI hype and AI slop? Well, as we know, many firms have
been jumping headfirst into trying to use AI for all manner of work
and it's widely believed that many of the massive layoffs of employees
now occurring are tied one way or another to these AI efforts, though
when asked some firms will insist AI isn't the cause. In any case,
much of the frantic AI efforts have been based on flat rate AI access
which you might view as "introductory" rates. But as the AI firms
moved to usage-sensitive pricing the fundamentals have drastically
changed.
Remember, the AI firms are desperate to pull in revenue to help offset
the enormous spending they've been doing on AI infrastructure such as
the massive, often polluting data centers that it seems almost
everyone despises irrespective of their political views on other
topics. And the AI firms have to push this even harder if they ever
hope to make a real profit on all this AI Slop.
On the other side of the coin, the firms trying to adopt AI are often
pushing their remaining employees to use it more and more, even
closely monitoring their usage to make sure they're using it enough to
satisfy management. This hasn't gone over well with many employees,
who may feel the AI systems aren't really contributing much of value
or may even be making their work more difficult. But since they're
being judged by how much they use AI, many employees have found ways
to -- shall we say -- "game" the system and show a great deal of
prized AI usage doing essentially worthless make-work tasks to pump up
their AI usage metrics.
As soon as you start paying for AI services on a usage sensitive basis
you're normally dealing in what's called "tokens", which are
essentially the usual pricing element for these systems. Tokens have
specific costs and depending on the specific kinds of AI tasks being
engaged, a very large number of tokens can be burned through very
quickly -- especially with agentic AI. This can actually result in
orders of magnitude more costs for using AI hitting firms quite
suddenly and unexpectedly. So now that 500 million dollars one month
AI bill starts to come into focus.
Combine the cost of AI tokens and the number of tokens being used by
employees of firms being urged (often against the employees' wills) to
embrace AI, and add in what can happen in the case that adequate
limits aren't preestablished for how many AI tokens employees can
actually use. SUR-PRISE!
But there's an even more important aspect to this. The executive
management of many firms -- not just the non-management employees --
including again at some very major firms -- are apparently starting to
question the actual value being obtained from their AI usage, even
when they're not faced with unexpected enormous AI usage bills.
There's increasing evidence that many firms are pulling back from even
their intended AI usage spends, because the actual results from AI are
reportedly simply not meeting their expectations.
And taken together, all these varied aspects of insane AI hype are
causing many observers to suspect that the so-called "AI bubble" is
perhaps heading toward bursting sooner rather than later, with all
that implies for possibly enormous damage to the broader economy. Even
if you personally choose to never voluntary use these AI systems, if
the AI bubble bursts, society as a whole is going to be affected and
the potentially catastrophic effects could roar through the global
economy like an enormous tidal wave. And if this happens, that 500
million dollar AI bill is going to look like relatively small potatoes
by comparison.
Even if regulators and governments get the AI firms on short leashes
immediately, the possibility of this kind of AI bubble triggered
economic crash is still unfortunately very real, and the negative
consequences -- except perhaps for the ultra-rich -- could be enormous
indeed.
- - -
L
- - -
--Lauren--
Lauren Weinstein
[email protected] (https://www.vortex.com/lauren)
Lauren's Blog: https://lauren.vortex.com
Mastodon: https://mastodon.laurenweinstein.org/@lauren
Signal: By request on need to know basis
Founder: Network Neutrality Squad: https://www.nnsquad.org
PRIVACY Forum: https://www.vortex.com/privacy-info
Co-Founder: People For Internet Responsibility
_______________________________________________
pfir mailing list
https://lists.pfir.org/mailman/listinfo/pfir
