The current OpenAI structure is bizarre and complex. It does important good things some of which this new arrangement will break. But the current structure also made OpenAI far less investable, which means giving away more of the company to profit maximizers, and causes a lot of real problems.
That was almost entirely good though, no? Myopic profit-maximization made OpenAI stray from the “beeline to AGI” suicidal course in favor of trying to R&D products people would buy, because if they did not do so, the money would dry up. If money instead comes from big investors buying into Sam Altman’s promises of making them god-kings, OpenAI is no longer “distracted” by needing to make their products short-term useful, and the actually existentially dangerous research (instead of “o3 lied to someone” kind of “dangerous”) is accelerated.
Simultaneously, the company’s safety culture is already dismantled beyond recovery, so there’s no reason to think they’d step up on safety just because the rat race’s pressures have lessened. Worst of both worlds.
IMO, profit caps making OpenAI less appealing for investment is/was the only actual feature of OpenAI’s structure that meaningfully hampers them; the rest is flavor text, particularly post-2023.
That was almost entirely good though, no? Myopic profit-maximization made OpenAI stray from the “beeline to AGI” suicidal course in favor of trying to R&D products people would buy, because if they did not do so, the money would dry up. If money instead comes from big investors buying into Sam Altman’s promises of making them god-kings, OpenAI is no longer “distracted” by needing to make their products short-term useful, and the actually existentially dangerous research (instead of “o3 lied to someone” kind of “dangerous”) is accelerated.
Simultaneously, the company’s safety culture is already dismantled beyond recovery, so there’s no reason to think they’d step up on safety just because the rat race’s pressures have lessened. Worst of both worlds.
IMO, profit caps making OpenAI less appealing for investment is/was the only actual feature of OpenAI’s structure that meaningfully hampers them; the rest is flavor text, particularly post-2023.