GPTs are being trained to predict text, not imitate humans. This task is actually harder than being human in many ways. You need to be smarter than the text generator to perfectly predict their output, and some text is the result of complex processes (e.g. scientific results, news) that even humans couldn't predict.
GPTs are solving a fundamentally different and often harder problem than just "be human-like". This means we shouldn't expect them to think like humans.
AI Alignment Posts
Popular Comments
Recent Discussion
Written as part of the AIXI agent foundations sequence, underlying research supported by the LTFF.
Epistemic status: In order to construct a centralized defense of AIXI I have given some criticisms less consideration here than they merit. Many arguments will be (or already are) expanded on in greater depth throughout the sequence. In hindsight, I think it may have been better to explore each objection in its own post and then write this post as a summary/centralized reference, rather than writing it in the middle of that process. Some of my takes have already become more nuanced. This should be treated as a living document.
With the possible exception of the learning-theoretic agenda, most major approaches to agent foundations research construct their own paradigm and mathematical tools which are...
My objection to this argument is that it not only assumes that Predictoria accepts it is plausibly being simulated by Adversaria, which seems like a pure complexity penalty over the baseline physics it would infer otherwise unless that helps to explain observations,
Let's assume for simplicity that both Predictoria and Adversaria are deterministic and nonbranching universes with the same laws of physics but potentially different starting conditions. Adversaria has colonized its universe and can run a trillion simulations of Predictoria in parallel. Again...
[EDIT: Never mind, this is just Kleene's second recursion theorem!]
Quick question about Kleene's recursion theorem:
Let's say F is a computable function from ℕ^N to ℕ. Is there a single computable function X from ℕ^N to ℕ such that
X = F(X, y_2,..., y_N) for all y_2,...,y_N in ℕ
(taking the X within F as the binary code of X in a fixed encoding) or do there need to be additional conditions?
This is an excerpt from the Introduction section to a book-length project that was kicked off as a response to the framing of the essay competition on the Automation of Wisdom and Philosophy. Many unrelated-seeming threads open in this post, that will come together by the end of the overall sequence.
If you don't like abstractness, the first few sections may be especially hard going.
Generalization
This sequence is a new story of generalization.
The usual story of progress in generalization, such as in a very general theory, is via the uncovering of deep laws. Distilling the real patterns, without any messy artefacts. Finding the necessities and universals, that can handle wide classes rather than being limited to particularities. The crisp, noncontingent abstractions. It is about opening black boxes. Articulating mind-independent, rigorous results, with no ambiguity and high...
Lots of interesting ideas here, but the connection to alignment still seems a bit vague.
Is misalignment really is a lack of sensitivity as opposed to a difference in goals or values? It seems to me that an unaligned ASI is extremely sensitive to context, just in the service of its own goals.
Then again, maybe you see Live Theory as being more about figuring out what the outer objective should look like (broad principles that are then localised to specific contexts) rather than about figuring out how to ensure an AI internalises specific values. And I can see potential advantages in this kind of indirect approach vs. trying to directly define or learn a universal objective.
TL;DR Having a good research track record is some evidence of good big-picture takes, but it's weak evidence. Strategic thinking is hard, and requires different skills. But people often conflate these skills, leading to excessive deference to researchers in the field, without evidence that that person is good at strategic thinking specifically. I certainly try to have good strategic takes, but it's hard, and you shouldn't assume I succeed!
Introduction
I often find myself giving talks or Q&As about mechanistic interpretability research. But inevitably, I'll get questions about the big picture: "What's the theory of change for interpretability?", "Is this really going to help with alignment?", "Does any of this matter if we can’t ensure all labs take alignment seriously?". And I think people take my answers to these...
A few points:
- Knowing a research field well makes it easier to assess how much other people know about it. For example, if you know ML, you sometimes notice that someone clearly doesn't know what they're talking about (or conversely, you become impressed by the fact that they clearly do know what they're talking about). This is helpful when deciding who to defer to.
- If you are a prominent researcher, you get more access to confidential/sensitive information and the time of prestigious people. This is true regardless of whether your strategic takes are good,
1Note: This post is intended to be the first in a broader series of posts about the difficult tradeoffs inherent in public access to powerful open source models. While this post highlights some dangers of open models and discusses the possibility of global regulation, I am not, in general, against open source AI, or supportive of regulation of open source AI today. On the contrary, I believe open source software is, in general, one of humanity’s most important and valuable public goods. My goal in writing this post is to call attention to the risks and challenges around open models now, so we can use the time we still have before risks become extreme, to collectively explore viable alternatives to regulation, if indeed such alternatives exist.
Background
Most research...
A few takes:
I believe that there is also an argument to be made that the AI safety community is currently very under-indexed on research into future scenarios where assumptions about the AI operator taking baseline safety precautions related to preventing loss of control do not hold.
I think you're mixing up two things: the extent to which we consider the possibility that AI operators will be very incautious, and the extent to which our technical research focuses on that possibility.
My research mostly focuses on techniques that an AI developer could us...
A few months ago, I accidentally used France as an example of a small country that it wouldn't be that catastrophic for AIs to take over, while giving a talk in France 😬