Why I think it's net harmful to do technical safety research at AGI labs

I think I agree that safety researchers should prefer not to take a purely ceremonial role at a big company if they have other good options, but I'm hesitant to conclude that no one should be willing to do it. I don't think it is remotely obvious that safety research at big companies is ceremonial.

There are a few reasons why some people might opt for a ceremonial role:

1. It is good for some AI safety researchers to have access to what is going on at top labs, even if they can't do anything about it. They can at least keep tabs on it and can use that experience later in their careers.

2. It seems bad to isolate capabilities researchers from safety concerns. I bet capabilities researchers would take safety concerns more seriously if they eat lunch every day with someone who is worried than if they only talk to each other.

3. If labs do engage in behavior that is flagrantly reckless, employees can act as whistleblowers. Non-employees can't. Even if they can't prevent a disaster, they can create a paper trail of internal concerns which could be valuable in the future.

4. Internal politics might change and it seems better to have people in place already thinking about these things.

Hi Conor,

Thank you.

I’m glad to see that you already linked to clarifications before. And that you gracefully took the feedback, and removed the prompt engineer role. I feel grateful for your openness here.

It makes me feel less like I’m hitting a brick wall. We can have more of a conversation.

~ ~ ~

The rest is addressed to people on the team, and not to you in particular:

There are grounded reasons why 80k’s approaches to recommending work at AGI labs – with the hope of steering their trajectory – has supported AI corporations to scale. While disabling efforts that may actually prevent AI-induced extinction.

This concerns work on your listed #1 most pressing problem. It is a crucial consideration that can flip your perceived total impact from positive to negative.

I noticed that 80k staff responses so far started by stating disagreement (with my view), or agreement (with a colleague’s view).

This doesn’t do discussion of it justice. It’s like responding to someone’s explicit reasons for concern that they must be “less optimistic about alignment”. This ends reasoned conversations, rather than opens them up.

Something I would like to see more of is individual 80k staff engaging with the reasoning.

A range of opinions from anonymous experts about the upsides and downsides of working on AI capabilities

I did read that compilation of advice, and responded to that in an email (16 May 2023):

"Dear [a],

People will drop in and look at job profiles without reading your other materials on the website. I'd suggest just writing a do-your-research cautionary line about OpenAI and Anthropic in the job descriptions itself.

Also suggest reviewing whether to trust advice on whether to take jobs that contribute to capability research.

• Particularly advice by nerdy researchers paid/funded by corporate tech. 
• Particularly by computer-minded researchers who might not be aware of the limitations of developing complicated control mechanisms to contain complex machine-environment feedback loops. 

Totally up to you of course.

Warm regards,

Remmelt"

 

We argue for this position extensively in my article on the topic

This is what the article says: 
"All that said, we think it’s crucial to take an enormous amount of care before working at an organisation that might be a huge force for harm. Overall, it’s complicated to assess whether it’s good to work at a leading AI lab — and it’ll vary from person to person, and role to role." 

So you are saying that people are making a decision about working for an AGI lab that might be (or actually is) a huge force for harm. And that whether it's good (or bad) to work at an AGI lab depends on the person – ie. people need to figure this out for them personally.

Yet you are openly advertising various jobs at AGI labs on the job board. People are clicking through and applying. Do you know how many read your article beforehand?

~ ~ ~
Even if they did read through the article, both the content and framing of the advice seems misguided. Noticing what is emphasised in your considerations. 

Here are the first sentences of each consideration section:
(ie. as what readers are most likely to read, and what you might most want to convey).

1. "We think that a leading — but careful — AI project could be a huge force for good, and crucial to preventing an AI-related catastrophe."
   • Is this your opinion about DeepMind, OpenAI and Anthropic? 
      
2. "Top AI labs are high-performing, rapidly growing organisations. In general, one of the best ways to gain career capital is to go and work with any high-performing team — you can just learn a huge amount about getting stuff done. They also have excellent reputations more widely. So you get the credential of saying you’ve worked in a leading lab, and you’ll also gain lots of dynamic, impressive connections."
   • Is this focussing on gaining prestige and (nepotistic) connections as an instrumental power move, with the hope of improving things later...?
   • Instead of on actually improving safety?
      
3. "We’d guess that, all else equal, we’d prefer that progress on AI capabilities was slower."
   • Why is only this part stated as a guess?  
     • I did not read "we'd guess that a leading but careful AI project, all else equal, could be a force of good". 
     • Or inversely:  "we think that continued scaling of AI capabilities could be a huge force of harm."
     • Notice how those framings come across very differently.
   • Wait, reading this section further is blowing my mind.
     • "But that’s not necessarily the case. There are reasons to think that advancing at least some kinds of AI capabilities could be beneficial. Here are a few"
     • "This distinction between ‘capabilities’ research and ‘safety’ research is extremely fuzzy, and we have a somewhat poor track record of predicting which areas of research will be beneficial for safety work in the future. This suggests that work that advances some (and perhaps many) kinds of capabilities faster may be useful for reducing risks."
       • Did you just argue for working on some capabilities because it might improve safety?  This is blowing my mind.
     • "Moving faster could reduce the risk that AI projects that are less cautious than the existing ones can enter the field."
       • Are you saying we should consider moving faster because there are people less cautious than us?  
       • Do you notice how a similarly flavoured argument can be used by and is probably being used by staff at three leading AGI labs that are all competing with each other? 
       • Did OpenAI moving fast with ChatGPT prevent Google from starting new AI projects?
     • "It’s possible that the later we develop transformative AI, the faster (and therefore more dangerously) everything will play out, because other currently-constraining factors (like the amount of compute available in the world) could continue to grow independently of technical progress."
       • How would compute grow independently of AI corporations deciding to scale up capability?
       • The AGI labs were buying up GPUs to the point of shortage. Nvidia was not able to supply them fast enough. How is that not getting Nvidia and other producers to increase production of GPUs?
       • More comments on the hardware overhang argument here.
     • "Lots of work that makes models more useful — and so could be classified as capabilities (for example, work to align existing large language models) — probably does so without increasing the risk of danger"
       • What is this claim based on?
          
4. "As far as we can tell, there are many roles at leading AI labs where the primary effects of the roles could be to reduce risks."
   1. As far as I can tell, this is not the case.
      1. For technical research roles, you can go by what I just posted. 
      2. For policy, I note that you wrote the following:
         "Labs also often don’t have enough staff... to figure out what they should be lobbying governments for (we’d guess that many of the top labs would lobby for things that reduce existential risks)."
         1. I guess that AI corporations use lobbyists for lobbying to open up markets for profit, and to not get actually restricted by regulations (maybe to move focus to somewhere hypothetically in the future, maybe to remove upstart competitors who can't deal with the extra compliance overhead, but don't restrict us now!).
         2. On prior, that is what you should expect, because that is what tech corporations do everywhere. We shouldn't expect on prior that AI corporations are benevolent entities that are not shaped by the forces of competition. That would be naive.

~ ~ ~
After that, there is a new section titled "How can you mitigate the downsides of this option?"

• That section reads as thoughtful and reasonable.
• How about on the job board, you link to that section in each AGI lab job description listed, just above the 'VIEW JOB DETAILS' button?  
  • For example, you could append and hyperlink 'Suggestions for mitigating downsides' to the organisational descriptions of Google DeepMind, OpenAI and Anthropic.
• That would help guide through potential applicants to AGI lab positions to think through their decision.
   

I would agree with Remmelt here. While upskilling people is helpful, if those people then go on to increase the rate of capabilities gain by AI companies, this is reducing the time the world has available to find solutions to alignment and AI regulation.

While, as a rule, I don't disagree with an industries increasing their capabilities, I do disagree with this when those capabilities knowingly lead to human extinction.

Thanks, I appreciate the paraphrase. Yes, that is a great summary.

 

I'm more optimistic e.g. that control turns out to be useful, or that there are hacky alignment techniques which work long enough to get through to the automation of crucial safety research

I hear this all the time, but I also notice that people saying it have not investigated the fundamental limits to controllability that you would encounter with any control system.

As a philosopher, would you not want to have a more generalisable and robust argument that this is actually going to work out?

I think that there are various non-research pathways for such people to (in expectation) increase the safety of the lab they're working at

I'm curious about the pathways you have in mind. I may have missed something here.

 

e.g. I think that one class of work people in labs could do is capabilities monitoring, and I think that if this were done to a good standard it could in fact help to reduce perceived-safety to outsiders in a timely fashion

I'm skeptical that that would work in this corporate context. 

"Capabilities" are just too useful economically and can creep up on you. Putting aside whether we can even measure comprehensively enough for "dangerous capabilities".

In the meantime, it's great marketing to clients, to the media, and to national interests:
You are working on AI systems that could become so capable, that you even have an entire team devoted to capabilities monitoring.
 
 

I guess I'm quite sceptical that signals like "well the safety team at this org doesn't really have any top-tier researchers and is generally a bit badly thought of" will be meaningfully legible to the relevant outsiders, so I don't really think that reducing the quality of their work will have too much impact on perceived safety

This is interesting. And a fair argument. Will think about this.

I would agree that this is a good summary:

Improving the quality/quantity of output from safety teams within AI labs has a (much) bigger impact on perceived safety of the lab than it does on actual safety of the lab. This is therefore the dominant term in the impact of the team's work. Right now it's negative.

If perception of safety is higher than actual safety, it will lead to underinvestment of future safety, which increases the probability of failure of the system.