OpenAI’s Political Proposals for AI Job Losses Have One Thing in Common – They Protect OpenAI

OpenAI’s thirteen-page Industrial Policy for the Intelligence Age, released on April 6, is a good example of the type of corporate document that comes wrapped in moral seriousness. It begins with language that someone might use to appear concerned but not overly so at a dinner party. It warns that the shift to superintelligence has already begun. Jobs will be eliminated. Industries will change. It will be felt by communities. The document then gradually starts to imply, paragraph by paragraph, that the cleanup should probably be done by someone else.

The framing has a historical bent. The Progressive Era. The New Deal. The definition of work was rewritten by labor protections developed after the combustion engine. It’s a sophisticated rhetorical device that isn’t wholly pessimistic. Previous disruptions did necessitate new labor-capital contracts. Reading it, though, gives the impression that OpenAI has positioned itself as the steam engine in this comparison rather than the factory owner. The main character is technology. The business is merely a passenger.

A closer look reveals a subtle familial resemblance between the proposals. Eventually, use democratic procedures to modernize the tax system. Increase access to healthcare. Reconsider your retirement funds. Although creating the rulebook from the incumbent’s desk is a form of entrenchment in and of itself, the document is careful to point out that common-sense AI regulation should not entrench incumbents. Almost nothing here requires OpenAI to take any particular action at its own expense in the near future. The requests are directed at Congress, state governments, and a potential future political agreement.

Will Manidis, a writer for Substack, expressed it more frankly than most journalists have. He wrote that OpenAI is suggesting that others think about eventually requesting that OpenAI pay more taxes via a democratic process. The line stings because it lands. Joshua Achiam, the company’s chief futurist, acknowledges the elephant in the room in the companion video: employees are reluctant to work together because they fear being replaced. He speaks the truth. The policy itself is less forthright in what follows.

The issue of timing is another. The Wall Street Journal revealed that OpenAI had fallen short of internal revenue and user growth goals three weeks after the policy was lifted. According to reports, CFO Sarah Friar expressed doubts about revenue’s ability to meet $600 billion in future compute commitments. A billion weekly active users had not been reached by ChatGPT. Oracle, Nvidia, Microsoft, CoreWeave, and SoftBank investors saw a decline in their stock. The Journal article was dubbed clickbait by a spokesman. He maintained that the company was operating at full capacity.

Perhaps it is. Perhaps it isn’t. The discrepancy between the company’s operational reality of struggling to keep enterprise revenue ahead of its energy bill and its declared belief that superintelligence is imminent enough to warrant rewriting the social contract is more difficult to ignore. The urgency changes if the disruption is that close. The policy appears to be preemptive coverage if it isn’t.

It’s difficult to ignore the fact that nearly all of the document’s recommendations make requests of the public sector and hardly any of OpenAI. According to the policy, data centers should cover their own energy costs. They shouldn’t be subsidized by households. That’s a fair principle. Additionally, OpenAI was able to commit to it without the need for a thirteen-page manifesto. For now, the remainder reads more like a meticulous brief from someone who already has excellent attorneys than it does like a New Deal.