There are now at least two AI development companies purporting to research safe AI and both have existed for a couple years now, so I think it's worth taking another look at how safe "safe" AI development is.
One of those companies is GoodAI. I don't know a lot about them beyond what's on their website. They seem clearly to be aware of AI safety concerns and the need for alignment but are also pursuing capabilities research. OpenAI is the other and is similarly pursuing capabilities research but, at least based on what I know about them publicly, only go so far as to say they want safe AI, although OpenAI does employ at least one person known to be actively working on AI safety. There may be other companies in the safe AI development space but to the best of my knowledge other efforts do not make explicit statements about safety and are only focused on capabilities (although AI for self-driving cars, for example, has certain mundane safety concerns different from those of AI safety research).
When OpenAI was started there was some discussion about whether it was a good idea. Ben Hoffman said no, Nate Soares had a positive reaction, and others had mixed responses. Some quick searching hasn't turned up any explicit opinions on GoodAI, so presumably people would feel similarly about them as they do about OpenAI. A quick rehash on some of the arguments:
We're a couple years on now, though, and given that AI seems to be on a strong upward swing, does it make sense to encourage more companies to be like OpenAI and GoodAI and target safe AI, perhaps via a self-regulatory organization, or is encouraging safety as an explicit goal in capabilities research unlikely to have much effect?
I'm inclined to suspect that some attention to safety is better than none because it gives a wedge with which to push for more safety later, so I'm especially curious about arguments that it wouldn't help.
I'm not sure I like the capabilities vs alignment frame. My view is: Alignment will be achieved through the right set of capabilities. A big part of FAI is figuring out which capabilities needed for FAI aren't needed for UFAI. If you could answer that question, then you've reduced FAI to an AI problem. And in order to answer it, you're going to want to spend a lot of time thinking about AI capabilities. Nick Bostrom's book is a heroic effort by someone who is not an AI capabilities expert to try to work on FAI, and it has a lot of interesting ideas. But as I learn more about AI, I start to see flaws in his thinking.
I'm not sure why OpenAI is working on deep reinforcement learning, though. Yes, it's trendy. But expert systems were also trendy once. If our "safe AI" groups just work on whatever is trendy at the moment, how are they different than non-"safe AI" groups? I'm currently hoping deep reinforcement learning doesn't go anywhere.
Maybe the meta question should be: what's the best way to influence which research areas are trendy? I think this talk was a positive development. I'm told there are now a decent number of researchers working on the "simple theorems, simple experiments" approach Rahimi advocates.
It also wouldn't surprise me if this is also a faster way to make progress in the long run. Ultimately, scientists reached transmutation before alchemists did. It's not entirely clear to me whether Rahimi style insight should be considered blessed "alignment" research or cursed "capabilities" research, but I'm leaning towards optimism.
If people want to do things that are unambiguously helpful, here's a different frame. Suppose we model the quality of an AI system as the product of the insight of the researchers, the amount of data they've got, and the amount of hardware they've got.
It's not totally clear to me when "insight" is positive or negative. But it seems likely to me that restricting hardware, so researchers are forced to use more insight instead of brute forcing things with black boxes, would be valuable. (Intuition: if the quality of a system is held constant, we would prefer for quality to be achieved because researchers have deep insight into the problem they're solving.) So maybe it'd be good to push for a global tax on GPUs or something. If a political movement forms around technological employment, they could agitate for this. Hardware is easier to regulate than software in any case.
Data is a bit more complex, because having a lot of training data for a task makes it easier to develop an AI to perform that task. So that suggests increasing the amount of data available for training AIs on ethics-related tasks, and decreasing the amount of data available for training AIs on other tasks. Not totally sure what this would look like.