The OpenAI debacle has reached its climax with a new board, the reinstatement of Sam Altman as CEO, and, well, whatever I write will be out of date by the time I hit send. You can read about some of the salient details in the New York Times. Much is still closely guarded.
These past 96 hours have mattered. This wasnât just about Silicon Valley intrigue (although for many on the West Coast, it seemed to be about that). Letâs pretend it was about the key questions that were the substance of my book, Exponential â how do we organise and govern a society in a time of fast technological change?
Everyone has a plan until they are punched in the mouth
Here is how I think about this.Â
It doesnât matter that OpenAI started as a non-profit and now wants to pursue profitable projects. Circumstances change. Facts change. It morphed over the past seven years. It suffered its own exponential gap between the prospects of deploying a useful enterprise technology (for that is what an LLM is) and the governance structure and mission it had in place.Â
Companies should not self-regulate. Society, through the sovereign state, should set the parameters under which firms operate. And firms should operate within those guardrails. It is just so much easier. The moment a firm serves two masters, someone will get the shaft. Google serves shareholders but through promises to advertisers and users. We get shanked with an increasingly poor search experience.Â
OpenAI was serving the âbenefit of humanityâ and customers. The âbenefit of humanityâ was cutesy when there were no real products and, frankly, nothing useful you could do with the technology. But it doesnât make sense when a firm actually has something useful to sell. As Logan Roy put it âMoney wins.â
(Iâll say more about the âbenefit of humanityâ momentarily.)
However, given the charter of the non-profit which owned the company was to act, in the boardâs determination, for the âbenefit of humanityâ, it doesnât seem to me that the board could not justify its actions. They could, potentially, have communicated more clearly as I argued. But the loophole in the charter was the power vested in the board against this waffly notion of the âbenefit of humanityâ. This isnât the world of R Daneel Olivawâs âzeroth law of roboticsâ. Itâs the real world.
More helpful than that is Milton Friedmanâs instructions for a firm. Now, Iâve criticised Friedmanâs narrow definition of a company many times. Itâs quite problematic. But it is more crisp, more understandable, than the benefit of humanityâ. With Friedmanâs definition, at least the scorpion recognises what it is. Somewhere in between the âindescribably greater goodâ and the mercenary firm is the organisation that is connected to its customers, community, shareholders, employees and finding a space between them.Â
Iâm sorry, Sam, I canât do that
The problem with the concept of âthe benefit of humanityâ is that, well, itâs incoherent at best.Â