Audited Hype
Read a Guardian piece by Samantha Oltman arguing that AI absolutism, the tendency to see the technology as either civilization-ending or civilization-saving, is itself a product being sold by the people who stand to profit from the panic. The argument is sensible as far as it goes. Dario Amodei predicting that AI is "a general labor substitute for humans" while also running the company that makes the labor substitute possible is a conflict of interest. Jensen Huang telling people they won't lose their jobs to AI but will lose them to someone who uses AI is the kind of formulation that sounds reassuring until you think a little deeper about that means for you or your children.
Oltman quotes Columbia economist Suresh Naidu and UC Berkeley's Martin Beraja to argue that the job displacement numbers are overblown, that tech layoffs had more to do with pandemic-era overhiring than with AI, that software is only 4 to 6% of GDP so Claude Code can't eat the whole economy. While all of this may be true, "the apocalyptic scenario hasn't happened yet" and "the apocalyptic scenario won't happen" are very different claims, but the piece treats them as interchangeable. Sam Altman walking back his predictions about entry-level job displacement is presented as evidence that the fears were wrong. It's equally evidence that the technology is slower than advertised, which is not the same thing knowing for a fact that it is harmless.
Gig workers being managed by algorithm, the gradual extension of that model to white-collar work, the use of AI to extract more productivity from people who are supposed to feel grateful for the extraction: are things that are already happening. Oltman mentions it and moves on to a hopeful paragraph about "a lot of different little AIs from little responsible players" and the possibility of a labor movement resurgence, neither of which she has any particular evidence for. The Industrial Revolution analogy at the end, that worker solidarity eventually won concessions even if it took time is an useful one.
I've worked in IT long enough to remember when "the internet changes everything" was also a claim being made by people who stood to profit from the panic and the enthusiasm in equal measure. Some of what was promised happened much of it didn't. What nobody could predict was which parts, or on what timeline, or who would be left holding the bag when the timeline slipped. The Guardian piece is right that AI absolutism is a marketing strategy. It's just not right that the corrective is a different set of confident predictions about where things are headed. Naidu is quoted saying there's no control group for what we're living through and that is a very honest assessment.
Oltman quotes Columbia economist Suresh Naidu and UC Berkeley's Martin Beraja to argue that the job displacement numbers are overblown, that tech layoffs had more to do with pandemic-era overhiring than with AI, that software is only 4 to 6% of GDP so Claude Code can't eat the whole economy. While all of this may be true, "the apocalyptic scenario hasn't happened yet" and "the apocalyptic scenario won't happen" are very different claims, but the piece treats them as interchangeable. Sam Altman walking back his predictions about entry-level job displacement is presented as evidence that the fears were wrong. It's equally evidence that the technology is slower than advertised, which is not the same thing knowing for a fact that it is harmless.
Gig workers being managed by algorithm, the gradual extension of that model to white-collar work, the use of AI to extract more productivity from people who are supposed to feel grateful for the extraction: are things that are already happening. Oltman mentions it and moves on to a hopeful paragraph about "a lot of different little AIs from little responsible players" and the possibility of a labor movement resurgence, neither of which she has any particular evidence for. The Industrial Revolution analogy at the end, that worker solidarity eventually won concessions even if it took time is an useful one.
I've worked in IT long enough to remember when "the internet changes everything" was also a claim being made by people who stood to profit from the panic and the enthusiasm in equal measure. Some of what was promised happened much of it didn't. What nobody could predict was which parts, or on what timeline, or who would be left holding the bag when the timeline slipped. The Guardian piece is right that AI absolutism is a marketing strategy. It's just not right that the corrective is a different set of confident predictions about where things are headed. Naidu is quoted saying there's no control group for what we're living through and that is a very honest assessment.
Filed under:
Technology