[NOW CLOSED]
MIRI Technical Governance Team is hiring, please apply and work with us!

We are looking to hire for the following roles:

• Technical Governance Researcher (2-4 hires)
• Writer (1 hire)

The roles are located in Berkeley, and we are ideally looking to hire people who can start ASAP. The team is currently Lisa Thiergart (team lead) and myself.

We will research and design technical aspects of regulation and policy that could lead to safer AI, focusing on methods that won’t break as we move towards smarter-than-human AI. We want to design policy that allows us to safely and objectively assess the risks from powerful AI, build consensus around the risks we face, and put in place measures to prevent catastrophic outcomes.

The team will likely work on:

• Limitations of current proposals such as RSPs
• Inputs into regulations, requests for comment by policy bodies (ex. NIST/US AISI, EU, UN)
• Researching and designing alternative Safety Standards, or amendments to existing proposals
• Communicating with and consulting for policymakers and governance organizations

If you have any questions, feel free to contact me on LW or at peter@intelligence.org 

I think the Claude Sonnet Golden Gate Bridge feature is not crispy aligned with the human concept of "Golden Gate Bridge". 
It brings up the San Fransisco fog far more than it would if it was just the bridge itself. I think it's probably more like Golden Gate Bridge + SF fog + a bunch of other things (some SF related, some not). 
This isn't particularly surprising, given these are related ideas (both SF things), and the features were trained in an unsupervised way. But still seems kinda important that the "natural" features that SAEs find are not like exactly intuitively natural human concepts.

• It might be interesting to look at how much the SAE training data actually mentions the fog and the Golden Gate Bridge together
• I don't really think that this is super important for "fragility of value"-type concerns, but probably is important for people who think we will easily be able to understand the features/internals of LLMs

Almost all of my Golden Gate Claude chats mention the fog. Here is a not particularly cherrypicked example:

↑ comment by gwern · 2024-05-24T20:32:15.798Z · LW(p) · GW(p)

I don't really think that this is super important for "fragility of value"-type concerns, but probably is important for people who think we will easily be able to understand the features/internals of LLMs

I'm not surprised if the features aren't 100% clean, because this is after all a preliminary research prototype of a small approximation of a medium-sized version of a still sub-AGI LLM.

But I am a little more concerned that this is the first I've seen anyone notice that the cherrypicked, single, chosen example of what is apparently a straightforward, familiar, concrete (literally) concept, which people have been playing with interactively for days, is clearly dirty and not actually a 'Golden Gate Bridge feature'. This suggests it is not hard to fool a lot of people with an 'interpretable feature' which is still quite far from the human concept. And if you believe that it's not super important for fragility-of-value because it'd have feasible fixes if noticed, how do you know anyone will notice?

Replies from: joel-burget, Charlie Steiner, kave, kromem

↑ comment by Joel Burget (joel-burget) · 2024-05-24T20:43:28.043Z · LW(p) · GW(p)

The Anthropic post itself said more or less the same:

↑ comment by Charlie Steiner · 2024-05-25T05:15:17.246Z · LW(p) · GW(p)

I'm not surprised if the features aren't 100% clean, because this is after all a preliminary research prototype of a small approximation of a medium-sized version of a still sub-AGI LLM.

It's more like a limitation of the paradigm, imo. If the "most golden gate" direction in activation-space and the "most SF fog" direction have high cosine similarity, there isn't a way to increase activation of one of them but not the other. And this isn't only a problem for outside interpreters - it's expensive for the AI's further layers to distinguish close-together vectors, so I'd expect the AI's further layers to do it as cheaply and unreliably as works on the training distribution, and not in some extra-robust way that generalizes to clamping features at 5x their observed maximum.

↑ comment by kave · 2024-05-24T21:39:29.265Z · LW(p) · GW(p)

FWIW, I had noticed the same but had thought it was overly split (“Golden Gate Bridge, particularly its fog, colour and endpoints”) rather than dirty.

↑ comment by kromem · 2024-05-25T10:19:34.178Z · LW(p) · GW(p)

While I think you're right it's not cleanly "a Golden Bridge feature," I strongly suspect it may be activating a more specific feature vector and not a less specific feature.

It looks like this is somewhat of a measurement problem with SAE. We are measuring SAE activations via text or image inputs, but what's activated in generations seems to be "sensations associated with the Golden gate bridge."

While googling "Golden Gate Bridge" might return the Wikipedia page, whats the relative volume in a very broad training set between encyclopedic writing about the Golden Gate Bridge and experiential writing on social media or in books and poems about the bridge?

The model was trained to complete those too, and in theory should have developed successful features for doing so.

In the research examples one of the matched images is a perspective shot from physically being on the bridge, a text example is talking about the color of it, another is seeing it in the sunset.

But these are all the feature activations when acting in a classifier role. That's what SAE is exploring - give it a set of inputs and see what lights it up.

Yet in the generative role this vector maximized keeps coming up over and over in the model with content from a sensory standpoint.

Maybe generation based on functional vector manipulations will prove to be a more powerful interpretability technique than SAE probing passive activations alone?

In the above chat when that "golden gate vector" is magnified, it keeps talking about either the sensations of being the bridge as if its physical body with wind and waves hitting it or the sensations of being on the bridge. It even generates towards the end in reflecting on the knowledge of the activation about how the sensations are overwhelming. Not reflecting on the Platonic form of an abstract concept of the bridge, but about overwhelming physical sensations of the bridge's materialism.

I'll be curious to see more generative data and samples from this variation, but it looks like generative exploration of features may offer considerably more fidelity to their underlying impact on the network than just SAE. Very exciting!!

Wow, what is going on with AI safety

Status: wow-what-is-going-on, is-everyone-insane, blurting [EA · GW], hope-I-don’t-regret-this

Ok, so I have recently been feeling something like “Wow, what is going on? We don’t know if anything is going to work, and we are barreling towards the precipice. Where are the adults in the room?”

People seem way too ok with the fact that we are pursuing technical agendas that we don’t know will work and if they don’t it might all be over. People who are doing politics/strategy/coordination stuff also don’t seem freaked out that they will be the only thing that saves the world when/if the attempts at technical solutions don’t work.

And maybe technical people are ok doing technical stuff because they think that the politics people will be able to stop everything when we need to. And the politics people think that the technical people are making good progress on a solution.

And maybe this is the case, and things will turn out fine. But I sure am not confident of that.

And also, obviously, being in a freaked out state all the time is probably not actually that conducive to doing the work that needs to be done.

Technical stuff

For most technical approaches to the alignment problem, we either just don’t know if they will work, or it seems extremely unlikely that they will be fast enough.

• Prosaic
  • We don’t understand the alignment problem well enough to even know if a lot of the prosaic solutions are the kind of thing that could work. But despite this, the labs are barreling on anyway in the hope that the bigger models will help us answer this question.
  • (Extremely ambitious) mechanistic interpretability seems like it could actually solve the alignment problem, if it succeeded spectacularly. But given the rate of capabilities progress, and the fact that the models only get bigger (and probably therefore more difficult to interpret), I don’t think mech interp will solve the problem in time.
    • Part of the problem is that we don’t know what the “algorithm for intelligence” is, or if such a thing even exists. And the current methods seem to basically require that you already know and understand the algorithm you’re looking for inside the model weights.
  • Scalable oversight seems like the main thing the labs are trying, and seems like the default plan for attempting to align the AGI. And we just don’t know if it is going to work. The part to scalable oversight solving the alignment problem seems to have multiple steps where we really hope it works, or that the AI generalizes correctly.
    • The results from the OpenAI critiques paper don’t seem all that hopeful. But I’m also fairly worried that this kind of toy scalable oversight research just doesn’t generalize.
    • Scalable oversight also seems like it gets wrecked if there are sharp capabilities jumps.
  • There are control/containment plans where you are trying to squeeze useful work out of a system that might be misaligned. I’m very glad that someone is doing this, and it seems like a good last resort. But also, wow, I am very scared that these will go wrong.
    • These are relying very hard on (human-designed) evals and containment mechanisms.
    • Your AI will also ask if it can do things in order to do the task (eg learn a new skill). It seems extremely hard to know which things you should and shouldn’t let the AI do.
• Conceptual, agent foundations (MIRI, etc)
  • I think I believe that this has a path to building aligned AGI. But also, I really feel like it doesn’t get there any time soon, and almost certainly not before the deep learning prosaic AGI is built. The field is basically at the stage of “trying to even understand what we’re playing with”, and not anywhere close to “here’s a path to a solution for how to actually build the aligned AGI”.

Governance (etc)

• People seem fairly scared to say what they actually believe.
  • Like, c’mon, the people building the AIs say that these might end the world. That is a pretty rock solid argument that (given sufficient coordination) they should stop. This seems like the kind of thing you should be able to say to policy makers, just explicitly conveying the views of the people trying to build the AGI.
    • (But also, yes, I do see how “AI scary” is right next too “AI powerful”, and we don’t want to be spreading low fidelity versions of this.)
• Evals
  • Evals seem pretty important for working out risks and communicating things to policy makers and the public.
  • I’m pretty worried about evals being too narrow, and so as long as the AI can’t build this specific bioweapon then it’s fine to release it into the world.
  • There is also the obvious question of “What do we do when our evals trigger?”. We need either sufficient coordination between the labs for them to stop, or for the government(s) to care enough to make the labs stop.
  • But also this seems crazy, like “We are building a world-changing, notoriously unpredictable technology, the next version or two might be existentially dangerous, but don’t worry, we’ll stop before it gets too dangerous.” How is this an acceptable state of affairs?
  • By default I expect RSPs to either be fairly toothless and not restrict things or basically stop people from building powerful AI at all (at which point the labs either modify the RSP to let them continue, or openly break the RSP commitment due to claimed lack of coordination)
    • For RSPs to work, we need stop mechanisms to kick in before we get the dangerous system, but we don’t know where that is. We are hoping that by iteratively building more and more powerful AI we will be able to work out where to stop.

I've seen a bunch of places where them people in the AI Optimism cluster dismiss arguments that use evolution as an analogy (for anything?) because they consider it debunked by Evolution provides no evidence for the sharp left turn [LW · GW]. I think many people (including myself) think that piece didn't at all fully debunk the use of evolution arguments when discussing misalignment risk. A people have written what I think are good responses to that piece; many of the comments, especially this one [LW(p) · GW(p)], and some [LW · GW] posts [LW · GW].

I don't really know what to do here. The arguments often look like:
A: "Here's an evolution analogy which I think backs up my claims."
B: "I think the evolution analogy has been debunked and I don't consider your argument to be valid."
A: "I disagree that the analogy has been debunked, and think evolutionary analogies are valid and useful". 

The AI Optimists seem reasonably unwilling to rehash the evolution analogy argument, because they consider this settled (I hope I'm not being uncharitable here). I think this is often a reasonable move, like I'm not particularly interested in arguing about climate change or flat-earth because I do consider these settled. But I do think that the evolution analogy argument is not settled. 

One might think that the obvious move here is to go to the object-level. But this would just be attempting to rehash the evolution analogy argument again; a thing that the AI Optimists seem (maybe reasonably) unwilling to do. 

Is there a name of the thing where an event happens and that specific event is somewhat surprising, but overall you would have expected something in that reference class (or level of surprisingness) to happen?

E.g. It was pretty surprising that Ivanka Trump tweeted about Situational Awareness, but I sort of expect events this surprising to happen.

My Favourite Slate Star Codex Posts

This is what I send people when they tell me they haven't read Slate Star Codex and don't know where to start.

Here are a bunch of lists of top ssc posts:

These lists are vaguely ranked in the order of how confident I am that they are good

https://guzey.com/favorite/slate-star-codex/

https://slatestarcodex.com/top-posts/

https://nothingismere.com/2015/09/12/library-of-scott-alexandria/

https://www.slatestarcodexabridged.com/ (if interested in psychology almost all the stuff here is good https://www.slatestarcodexabridged.com/Psychiatry-Psychology-And-Neurology)

https://danwahl.net/blog/slate-star-codex

I think that there are probably podcast episodes of all of these posts listed below. The headings are not ranked by which heading I think is important, but within each heading they are roughly ranked. These are missing any new great ACX posts. I have also left off all posts about drugs, but these are also extremely interesting if you like just reading about drugs, and were what I first read of SSC. 

If you are struggling to pick where to start I recommend either Epistemic Learned Helplessness or Nobody is Perfect, Everything is Commensurable. 

The ‘Thinking good’ posts I think have definitely improved how I reason, form beliefs, and think about things in general. The ‘World is broke’ posts have had a large effect on how I see the world working, and what worlds we should be aiming for. The ‘Fiction’ posts are just really really good fiction short stories. The ‘Feeling ok about yourself’ have been extremely helpful for developing some self-compassion, and also for being compassionate and non-judgemental about others; I think these posts specifically have made me a better person.

Posts I love:

Thinking good

https://slatestarcodex.com/2019/06/03/repost-epistemic-learned-helplessness/

https://slatestarcodex.com/2014/08/10/getting-eulered/

https://slatestarcodex.com/2013/04/13/proving-too-much/

https://slatestarcodex.com/2014/04/15/the-cowpox-of-doubt/

World is broke and how to deal with it

https://slatestarcodex.com/2014/07/30/meditations-on-moloch/

https://slatestarcodex.com/2014/12/19/nobody-is-perfect-everything-is-commensurable/

https://slatestarcodex.com/2013/07/17/who-by-very-slow-decay/

https://slatestarcodex.com/2014/02/23/in-favor-of-niceness-community-and-civilization/
 

Fiction

https://slatestarcodex.com/2018/10/30/sort-by-controversial/

https://slatestarcodex.com/2015/04/21/universal-love-said-the-cactus-person/

https://slatestarcodex.com/2015/08/17/the-goddess-of-everything-else-2/ (read after moloch stuff I think)

Feeling ok about yourself

https://slatestarcodex.com/2015/01/31/the-parable-of-the-talents/

https://slatestarcodex.com/2014/08/16/burdens/

Other??

https://slatestarcodex.com/2014/09/10/society-is-fixed-biology-is-mutable/

https://slatestarcodex.com/2018/04/01/the-hour-i-first-believed/ (this is actual crazy, read last)