It's AI Week! [newsletter]
Surprise, it's AI week! And this newsletter is a short one. Enjoy!
(I)
Since OpenAI has been playing Game of Thrones for the last few days, I have AI top of mind. (For those wanting to catch up, Hardfork has been doing a pretty good job of covering the shenanigans surrounding Sam Altman's role at the company in case you want to catch up, but even they're not quite caught up yet I think?)
What happened? Apparently OpenAI's board and Sam Altman got into a power struggle, the board fired Altman. His co-founder left along with Altman in solidarity, and a bunch of employees threatened to quit as well. (It didn't come to that.) Altman, well-connected in the Valley as he is, got a huge outpouring of support on the socials as well. Then — plot twist! — Microsoft, OpenAI's biggest investor, hired Altman to run an independent AI research lap. So ended season one with quite the teaser. Cut to the next day, the launch of season 2 - where OpenAI reinstated Altman. That's roughly where we are.
Now, the background to all this is the structure of OpenAI: A non-profit that owns the for-profit company, similarly to say Mozilla (where the foundation also owns the company) with the idea that this keeps pure profit motivations out of the work. It's a safeguard against the lesser instincts, if I may phrase it a little dramatically. The folks installed at the board are there as safeguards against just that, and they lost big time.
Where does that leave us? It's too early to tell, but from what I can see right now it's likely that this will clear the way for OpenAI to become one of the more influential tech players (and effectively to join Big Tech soon-ish) and also, simultaneously be bad news for AI safety.
(II)
Shout out to Carla Hustedt who ran a great (and beautifully controversial) session about AI regulation at the German government's digital transformation summit, Digitalgipfel. The conversation is in German, though Youtube's real-time auto-translated subtitles have become really quite good. The debate starts around the 1:09 mark. For transparency reasons I should add that Carla (or rather her employer Stiftung Mercator) isn't just a client of mine but also a good friend who I deeply respect and I'm a real fan of her work, and this and the next item show why.
(III)
Speaking of which, German vice chancellor Robert Habeck was a participant in that debate mentioned above. In it, he insisted on not regulating foundation models and instead for industry self-regulation, which most experts agree is a bad idea as self-regulation in tech so far has shown us over and over. I truly think he's wrong on this one.
(IV)
In a shared post, Mark Surman & Camille François argue for "A Third Way on AI" and I quite like it. I've been following both their work since I was a senior fellow at Mozilla a few years back; Mark was Mozilla's CEO then, Camille was a fellow around the same time. This is good!
Brevity is the soul of narrative, so I'll leave you to it. Talk soon!
P.