Hastily written; may edit later

Thanks for mentioning this, Jan! We'd be happy to hear suggestions for additional judges. Feel free to email us at akash@alignmentawards.com and olivia@alignmentawards.com.  

Some additional thoughts:

1. We chose judges primarily based on their expertise and (our perception of) their ability to evaluate submissions about goal misgeneralization and corrigibility. Lauro, Richard, Nate, and John ade some of few researchers who have thought substantially about these problems. In particular, Lauro first-authored the first paper about goal misgeneralization and Nate first-authored a foundational paper about corrigibility.
2. We think the judges do have some reasonable credentials (e.g., Richard works at OpenAI, Lauro is a PhD student at the University of Cambridge, Nate Soares is the Executive Director of a research organization & he has an h-index of 12, as well as 500+ citations). I think the contest meets the bar of "having reasonably well-credentialed judges" but doesn't meet the bar of "having extremely well-credentialed judges (e.g., well-established professors with thousands of citations). I think that's fine.
3. We got feedback from several ML people before launching. We didn't get feedback that this looks "extremely weird" (though I'll note that research competitions in general are pretty unusual). 
4. I think it's plausible that some people will find this extremely weird (especially people who judge things primarily based on the cumulative prestige of the associated parties & don't think that OpenAI/500 citations/Cambridge are enough), but I don't expect this to be a common reaction.

Some clarifications + quick thoughts on Sam’s points:

1. The contest isn’t aimed primarily/exclusively at established ML researchers (though we are excited to receive submissions from any ML researchers who wish to participate). 
2. We didn’t optimize our contest to attract established researchers. Our contests are optimized to take questions that we think are at the core of alignment research and present them in a (somewhat less vague) format that gets more people to think about them.
3. We’re excited that other groups are running contests that are designed to attract established researchers & present different research questions. 
4. All else equal, we think that precise/quantifiable grading criteria & a diverse panel of reviewers are preferable. However, in our view, many of the core problems in alignment (including goal misgeneralization and corrigibility) have not been sufficiently well-operationalized to have precise/quantifiable grading criteria at this stage.

OK, Below I will provide links to few mathematically precise papers about AGI corrigibility solutions, with some comments. I do not have enough time to write short comments, so I wrote longer ones.

This list or links below is not a complete literature overview. I did a comprehensive literature search on corrigibility back in 2019 trying to find all mathematical papers of interest, but have not done so since.

I wrote some of the papers below, and have read all the rest of them. I am not linking to any papers I heard about but did not read (yet).

Math-based work on corrigibility solutions typically starts with formalizing corrigibility, or a sub-component of corrigibility, as a mathematical property we want an agent to have. It then constructs such an agent with enough detail to show that this property is indeed correctly there, or at least there during some part of the agent lifetime, or there under some boundary assumptions.

Not all of the papers below have actual mathematical proofs in them, some of them show correctness by construction. Correctness by construction is superior to having to have proofs: if you have correctness by construction, your notation will usually be much more revealing about what is really going on than if you need proofs.

Here is the list, with the bold headings describing different approaches to corrigibility.

Indifference to being switched off, or to reward function updates

Motivated Value Selection for Artificial Agents introduces Armstrong's indifference methods for creating corrigibility. It has some proofs, but does not completely work out the math of the solution to a this-is-how-to-implement-it level.

Corrigibility tried to work out the how-to-implement-it details of the paper above but famously failed to do so, and has proofs showing that it failed to do so. This paper somehow launched the myth that corrigibility is super-hard.

AGI Agent Safety by Iteratively Improving the Utility Function does work out all the how-to-implement-it details of Armstrong's indifference methods, with proofs. It also goes into the epistemology of the connection between correctness proofs in models and safety claims for real-world implementations.

Counterfactual Planning in AGI Systems introduces a different and more easy to interpret way for constructing a a corrigible agent, and agent that happens to be equivalent to agents that can be constructed with Armstrong's indifference methods. This paper has proof-by-construction type of math.

Corrigibility with Utility Preservation has a bunch of proofs about agents capable of more self-modification than those in Counterfactual Planning. As the author, I do not recommend you read this paper first, or maybe even at all. Read Counterfactual Planning first.

Safely Interruptible Agents has yet another take on, or re-interpretation of, Armstrong's indifference methods. Its title and presentation somewhat de-emphasize the fact that it is about corrigibility, by never even discussing the construction of the interruption mechanism. The paper is also less clearly about AGI-level corrigibility.

How RL Agents Behave When Their Actions Are Modified is another contribution in this space. Again this is less clearly about AGI.

Agents that stop to ask a supervisor when unsure

A completely different approach to corrigibility, based on a somewhat different definition of what it means to be corrigible, is to construct an agent that automatically stops and asks a supervisor for instructions when it encounters a situation or decision it is unsure about. Such a design would be corrigible by construction, for certain values of corrigibility. The last two papers above can be interpreted as disclosing ML designs that also applicable in the context of this stop when unsure idea.

Asymptotically unambitious artificial general intelligence is a paper that derives some probabilistic bounds on what can go wrong regardless, bounds on the case where the stop-and-ask-the-supervisor mechanism does not trigger. This paper is more clearly about the AGI case, presenting a very general definition of ML.

Anything about model-based reinforcement learning

I have yet to write a paper that emphasizes this point, but most model-based reinforcement learning algorithms produce a corrigible agent, in the sense that they approximate the ITC counterfactual planner from the counterfactual planning paper above.

Now, consider a definition of corrigibility where incompetent agents (or less inner-aligned agents, to use a term often used here) are less corrigible because they may end up damaging themselves, their stop buttons. or their operator by being incompetent. In this case, every convergence-to-optimal-policy proof for a model-based RL algorithm can be read as a proof that its agent will be increasingly corrigible under learning.

CIRL

Cooperative Inverse Reinforcement Learning and The Off-Switch Game present yet another corrigibility method with enough math to see how you might implement it. This is the method that Stuart Russell reviews in Human Compatible. CIRL has a drawback, in that the agent becomes less corrigible as it learns more, so CIRL is not generally considered to be a full AGI-level corrigibility solution, not even by the original authors of the papers. The CIRL drawback can be fixed in various ways, for example by not letting the agent learn too much. But curiously, there is very little followup work from the authors of the above papers, or from anybody else I know of, that explores this kind of thing.

Commanding the agent to be corrigible

If you have an infinitely competent superintelligence that you can give verbal commands to that it will absolutely obey, then giving it the command to turn itself into a corrigible agent will trivially produce a corrigible agent by construction.

Giving the same command to a not infinitely competent and obedient agent may give you a huge number of problems instead of course. This has sparked endless non-mathematical speculation, but in I cannot think of a mathematical paper about this that I would recommend.

AIs that are corrigible because they are not agents

Plenty of work on this. One notable analysis of extending this idea to AGI-level prediction, and considering how it might produce non-corrigibility anyway, is the work on counterfactual oracles. If you want to see a mathematically unambiguous presentation of this, with some further references, look for the section on counterfactual oracles in the Counterfactual Planning paper above.

Myopia

Myopia can also be considered to be feature that creates or improves or corrigibility. Many real-world non-AGI agents and predictive systems are myopic by construction: either myopic in time, in space, or in other ways. Again, if you want to see this type of myopia by construction in a mathematically well-defined way when applied to AGI-level ML, you can look at the Counterfactual Planning paper.

ETA: Koen recommends reading Counterfactual Planning in AGI Systems before (or instead of) Corrigibility with Utility Preservation [AF · GW]

Update: I started reading your paper "Corrigibility with Utility Preservation [AF · GW]".^[1]  My guess is that readers strapped for time should read {abstract, section 2, section 4} then skip to section 6.  AFAICT, section 5 is just setting up the standard utility-maximization framework and defining "superintelligent" as "optimal utility maximizer".

Quick thoughts after reading less than half:

AFAICT,^[2] this is a mathematical solution to corrigibility in a toy problem, and not a solution to corrigibility in real systems.  Nonetheless, it's a big deal if you have in fact solved the utility-function-land version which MIRI failed to solve.^[3]  Looking to applicability, it may be helpful for you to spell out the ML analog to your solution (or point us to the relevant section in the paper if it exists).  In my view, the hard part of the alignment problem is deeply tied up with the complexities of the {training procedure --> model} map, and a nice theoretical utility function is neither sufficient nor strictly necessary for alignment (though it could still be useful).

So looking at your claim that "the technical problem [is] mostly solved", this may or may not be true for the narrow sense (like "corrigibility as a theoretical outer-objective problem in formally-specified environments"), but seems false and misleading for the broader practical sense ("knowing how to make an AGI corrigible in real life").^[4]

Less important, but I wonder if the authors of Soares et al agree with your remark in this excerpt^[5]:

"In particular, [Soares et al] uses a Platonic agent model [where the physics of the universe cannot modify the agent's decision procedure]  to study a design for a corrigible agent, and concludes that the design considered does not meet the desiderata, because the agent shows no incentive to preserve its shutdown behavior. Part of this conclusion is due to the use of a Platonic agent model." 

1. ^{^}

   Btw, your writing is admirably concrete and clear.

   Errata:  Subscripts seem to broken on page 9, which significantly hurts readability of the equations.  Also there is a double-typo "I this paper, we the running example of a toy universe" on page 4.

2. ^{^}

   Assuming the idea is correct

3. ^{^}

   Do you have an account of why MIRI's supposed impossibility results (I think these exist?) are false?

4. ^{^}

   I'm not necessarily accusing you of any error (if the contest is fixated on the utility function version), but it was misleading to me as someone who read your comment but not the contest details.

5. ^{^}

   Portions in [brackets] are insertions/replacements by me

Hi Akash! Thanks for the quick clarifications, these make the contest look less weird and more useful than just a 500 word essay contest.

My feedback here is that I definitely got the 500 word essay contest vibe when I read the 'how it works' list on the contest home page, and this vibe only got reinforced when I clicked on the official rules link and skimmed the document there. I recommend that you edit the 'how it works' list to on the home page, to make it it much more explicit that the essay submission is often only the first step of participating, a step that will lead to direct feedback, and to clarify that you expect that most of the prize money will go to participants who have produced significant research beyond the initial essay. If that is indeed how you want to run things.

On judging: OK I'll e-mail you.

I have to think more about your question about posting a writeup on this site about what I think are the strongest proposals for corrigibility. My earlier overview writeup [LW · GW] that explored the different ways how people define corrigibility took me a lot of time to write, so there is an opportunity cost I am concerned about. I am more of an academic paper writing type of alignment researcher than a blogging all of my opinions on everything type of alignment researcher.

On the strongest policy proposal towards alignment and corrigibility, not technical proposal: if I limit myself to the West (I have not looked deeply into China, for example) then I consider the EU AI Act initiative by the EU to be the current strongest policy proposal around. It is not the best proposal possible, and there are a lot of concerns about it, but if I have to estimate expected positive impact among different proposals and initiatives, this is the strongest one.