"I share the concern that people working on valu..."


by Paul Christiano Jun 29 2016 updated Jun 29 2016

I share the concern that people working on value alignment won't understand what has been done before, or recognize e.g. MIRI's competencies, and so will reinvent the wheel (or, worse, fail to reinvent the wheel).

I think this post (and MIRI's communication more broadly) run a serious risk of seeming condescending. I don't think it matters much in the context of this post but I do think it matters more broadly. Also, I am concerned that MIRI will fail to realize that the research community really knows quite a lot more than MIRI about how to do good research, and so will be dismissive of mainstream views about how to do good research. In some sense the situation is symmetrical, and I think the best outcome is for everyone to recognize each other's expertise and treat each other with respect. (And I think that each side tends to ignore the other because the other ignores the one, and its obvious to both sides that the other side is doing badly because of it.)

In particular, it is very unclear whether people working on value alignment today are using the correct background assumptions, language, and division of problems are good. I don't think it's the case that new results should be able to be couched in the terms of the existing discussion.

So while this post might be appropriate for a random person discussing value alignment on LessWrong, I suspect it is inappropriate for a random ML researcher encountering the topic for the first time. Even if they have incorrect ideas about how to approach the value alignment problem, I think that seeing this would not help.

At the object level: I'm not convinced by the headline recommendation of the piece. I think that many plausible attacks on the problem (e.g. IRL, act-based agents) are going to look more like "solving everything at once" than like addressing one of the subproblems that you see as important. Of course those approaches will themselves be made out of pieces, but the pieces don't line up with the pieces in the current breakdown.

To illustrate the point, consider a community that thinks about AI safety from the perspective of what could go wrong. They say: "well, a robot could physically harm a person, or the robot could offend someone, or the robot could steal stuff…" You come to them and say "What you need is a better account of logical uncertainty." They respond: "What?" You respond by laying out a long agenda which in your view solves value alignment. The person responds: "OK, so you have some elaborate agenda which you think solves all of our problems. I have no idea what to make of that. How about you start by solving a subproblem, like preventing robots from physically injuring people?"

I know this is an unfair comparison, but I hope it helps illustrate how someone might feel about the current situation. I think it's worthwhile to try to get people to understand the thinking that has been done and build on it, but I think that it's important to be careful about alienating people unnecessarily and being unjustifiably condescending about it. Saying things like "most of the respect in the field…" also seems like it's just going to make things worse, especially when talking to people who have significantly more academic credibility than almost everyone anyone involved in AI control research.

Incidentally, the format of Arbital currently seems to exacerbate this kind of difficulty. It seems like the site is intended to uncover the truth (and e.g. posts are not labelled by author, they are just presented as consensus). A side effect is that if a post annoys me, it's not really clear what to do other than to be annoyed at Arbital itself.


Eliezer Yudkowsky

I agree this page is problematic in present form and probably needs to be rewritten by Rob Bensinger. As it stands, I suppose, it's trying too hard to rescue the sort of person who does try to solve the whole problem in 15 seconds using one simple trick - a class which unfortunately includes a number of prestigious people whose brain goes into some kind of… mode.

I see a lot of force in your worry here, but I don't know what I can do about it myself, except maybe rewrite the page so that it pretends to only be addressing outright cranks. I've made a couple of edits to that end, and to remove some of the particular language you singled out as problematic.

If it were up to you, how would you handle the problem of the 15-Second-Solvers, possibly ones with high status? Just give up and accept that we shall always have them with us?

Separately: I do worry that act-based agency is overreaching in how much it claims to solve with how simple of an idea. But it's not like you're putting it in uncritiqueably vague or delayed form, and it's not like you came up with it in 15 seconds (so far as I know). Building a persistent edifice and defending it in detail is fine.

The current technical agenda isn't supposed to be a complete solution to 'building AGI without getting killed'. Why would it be? Did something give you that impression?

Separately: re: Arbital: As of June 2016, we're not currently focused on debate features (like the probability bars) and are just trying to make Arbital useful for conveying explanations. Explanation does seem to be a subset of debate features-wise, but it also means that Arbital's main use case is for explaining knowledge that's already there. Which is why the public face of Arbital is focusing on math where things are less controversial. In other words, all I can do is bow my head and say 'sorry' about the fact that at present, all we can do is have different people write different Arbital pages and try to courteously link opposing views where they've been written up.