[dead]
> There was nothing ready or thought through.
This was not my view being somewhat close to the technologies there. Lamda did exist internally for several years, with pretty good demos.
It was clearly not GPT-4, but there was a lot of exec-level fear about releasing these systems and having them say something offensive.
"Responsibility" is kind of bullshit, but fear of bad press is very real.
I wouldn't be so sure, in my case ChatGPT passes the bar of being mildly useful but Bard is still absolutely useless. I can see two equally likely explanations for this: they simply can't manage to pull it off due to their culture or they can't release something that isn't massively more nerfed than the competition.