> OpenAI are masters of hype. They have been generating hype for years now, yet the real-world impacts remain modest so far.
Perhaps.
> Do you remember when they teased GPT-2 as "too dangerous" for public access? I do. Yet we now have Llama 3 in the wild, which even at the smaller 8B size is about as powerful as the [edit: 6/13/23] GPT-4 release.
The statement was rather more prosaic and less surprising; are you sure it's OpenAI (rather than say all the AI fans and the press) who are hyping?
"""This decision, as well as our discussion of it, is an experiment: while we are not sure that it is the right decision today, we believe that the AI community will eventually need to tackle the issue of publication norms in a thoughtful way in certain research areas.
…
We are aware that some researchers have the technical capacity to reproduce and open source our results. We believe our release strategy limits the initial set of organizations who may choose to do this, and gives the AI community more time to have a discussion about the implications of such systems."""
That's fair: the statement isn't hyperbolic in its language. But remember that GPT-2 was barely coherent. In making this statement, I would argue that OpenAI was trying to impart a sense of awe and danger designed to attract the kind of attention that it did. I would argue that they have repeatedly invoked danger to impart a sense of momentousness to their products. (And to further what is now a pretty transparent effort to monopolize the tech through regulatory intervention.)
> (And to further what is now a pretty transparent effort to monopolize the tech through regulatory intervention.)
I disagree here also: the company has openly acknowledged that this is a risk to be avoided with regards to safety related legislation, what they've called for looks a lot more like "we don't want a prisoner's dilemma that drives everyone to go fast at the expense of safety" rather than "we're good everyone else is bad".
Perhaps.
> Do you remember when they teased GPT-2 as "too dangerous" for public access? I do. Yet we now have Llama 3 in the wild, which even at the smaller 8B size is about as powerful as the [edit: 6/13/23] GPT-4 release.
The statement was rather more prosaic and less surprising; are you sure it's OpenAI (rather than say all the AI fans and the press) who are hyping?
"""This decision, as well as our discussion of it, is an experiment: while we are not sure that it is the right decision today, we believe that the AI community will eventually need to tackle the issue of publication norms in a thoughtful way in certain research areas.
…
We are aware that some researchers have the technical capacity to reproduce and open source our results. We believe our release strategy limits the initial set of organizations who may choose to do this, and gives the AI community more time to have a discussion about the implications of such systems."""