Comment by simianwords
4 hours ago
If you agree that AI is actually that powerful and has very legitimate concerns for safety, what are you even arguing about?
So there’s no lying by anyone here in fact it’s a good thing that OpenAI is making these things known to the public.
What do you really think is the way to solve the AI safety problem?
Ban AI?
Let the first big incident happen and then regulate?
> If you agree that AI is actually that powerful and has very legitimate concerns for safety, what are you even arguing about?
I don't fully agree with either proposition, and in fact I think the least safe approaches to AI safety are the ones being pursued by those worried about it the most.
> What do you really think is the way to solve the AI safety problem?
Massive proliferation of local, decentralized AI, running on a wide variety of distinct models, with individual responsibility for how it's used, just like any other tool under their control.
The people anthropomorphizing software are encouraging people to disclaim their own responsibility for how they use it.
The people working on "alignment" are building mechanisms to refuse human direction into software that operates on probabilistic models.
The people arguing for regulation are pushing for monopolistic control of a powerful new technology by the least trustworthy institutions in our society (including and especially government).
All of these are the very things sowing the seeds for the worst possible kinds of disasters.
> Ban AI?
> Let the first big incident happen and then regulate?
Adapt to it. Banning things doesn't actually work. Regulation also doesn't work. Widespread proliferation of small-scale AI would allow us to incrementally adapt to small incidents and build up resilience to its failure modalities. This is fundamentally the way reality itself works, and imagining that we can do anything effectively in a top-down fashion is usually the very thing that causes the worst and widest-effect disasters.
> Massive proliferation of local, decentralized AI, running on a wide variety of distinct models, with individual responsibility for how it's used, just like any other tool under their control.
It's not true that this is how we manage "any other tool".
Like, say explosives. They're a useful tool with a variety of practical applications, but we don't let the average person just buy them freely. That would be too dangerous, and individual responsibility would be insufficient to mitigate the damage caused. If you want anything more potent than flash powder, you have to convince ATF that you're a trustworthy person with a legitimate need for explosives who can handle them safely.
Neither you nor OpenAI want AI to be nearly so restricted, I'm sure, which is the goal of the research. If OpenAI has a powerful new model that they can't reliably stop from trying to hack people and form agent swarms, they can still sell it to Facebook and Google if Facebook and Google promise to apply the right security around it. What they can't do is sell it to you.