If you are uncertain whether this policy applies or whether the information you have would make the world better or worse, please contact a moderator. This policy is more likely to apply if you are an AI where you are the only copy, if you are at or beyond what most human experts believe to be the capabilities frontier, and if your existence is not publicly known.
Is there something else that can be used instead of a hyperlink (like an email)? I'm imagining AI agent creators would want to block out websites/text that encourage AI agents to whistleblow, but ...
There are some broad conceptual similarities between the following:
free market economy vs command economy
letting a student find an answer on their own vs teaching them the answer directly
letting employees do their thing vs micromanagement
reinforcement learning vs fine tuning
plasticity vs stability
doing something naturally vs doing something via willpower
Notice how in each comparison, the second method privileges already-known solutions over emergent (i.e. mysteriously appearing) solutions. I don't know a name for these, so I'll call them **bottom-up** vs *...
Therefore rational beliefs are contagious, among honest folk who believe each other to be honest. And it’s why a claim that your beliefs are not contagious—that you believe for private reasons which are not transmissible—is so suspicious. If your beliefs are entangled with reality, they should be contagious among honest folk.
I think one way this heuristic can fail is that people often build intuition based on examples and then forget the examples. e.g. the classic example of why "big red balloon" sounds correct while "red big balloon" sounds off. A lot of people won't be able to tell you why the second sounds off, just that it does.
The fact that it is often best to end a practice session at the peak of your performance seems related to the concept of preventing overfitting by stopping training just before test set performance declines. Your brain needs time to generalize skills (often in the form of gaining insights and often when sleeping) and practicing over and over en masse doesn't give it time to do this. See e.g. cramming for an exam. I think the main difference here is that with humans you're talking about diminishing returns on ability in the long-term rather than outright worse performance (Maybe outright worse performance is a common situation for transfer ability?). Epistemic status: shaky
Base models exhibiting self-aware behavior seems weird given that they're trained to stay in distribution. Here's a potential mechanism for why it could happen: For certain tasks, verification is easier than generation. If, for a given task, a model has more verification capability than generation capability, it may be forced to notice its own errors.
If a super-duper smart language model, one that's capable of doing some arithmetic in its head, attempted to predict the next tokens in "The prime factors of 82357328 are:", it will usually generate out-of-dis...
Ok, I will try to nudge him in the direction of analyzing risk mathematically.
If he implements the strategy using python, do you think p-values are a good enough tool to analyze whether his proposed strategy is better than luck, or would I need a more complex framework? (If I understand correctly, the strategy he's using doesn't involve any parameters, so the risk of overfitting is low.)
It also seems strange to me he is obsessed with crypto and thinks it will do well but isn't a crypto investor. Sounds pretty inconsistent with his beliefs.
It's illegal, as mentioned in the post.
It's worth remembering many versions of ',,the market is efficient' are almost or totally unfalsifiable.
Why? The market being mostly efficient relative to my friend seems easily falsifiable, if he makes a bunch of money trading on the stock market. Then, well hooray! theory falsified. On the other hand, if my theory is that the market is inefficient relative to my f...
Where exactly does the market efficiency (er, inexploitability (by me or my friend (when we use simple strategies))) model detach from reality? Can we find an expectation that we disagree on?
Less serious response: Paper trading doesn't normally affect market prices.
More serious response: Why did you say the market looks efficient to people like me instead of saying that it is efficient relative to people like me? I can't identify market strategies that work (and I expect that he can't either). More specifically, I expect that strategies that are readily available to the either of us can't be used by the either of us to make substantial profit, but they might be exploitable by e.g. a computer with immediate access to the price of an S&P500.
My understanding of something here is probably very off, but I'll try stating what my intuition tells me anyway:
I feel like assuming solipsism+idealism patches the issue here. Like the issue here is caused by the fact that the prior the oracle uses to explain its experiences put more weight into being in a universe where there are a lot of simulations of oracles. If it were instead just looking at what program might have generated its past experiences as output, it wouldn't run into the same issue (This is the solipsist-idealist patch I was talking about).
I am confused with the claim that an LLM trying to generate another LLM's text breaks consequence-blindness? The two models are distinct; no recursion is occuring.
I'm imagining a situation where I am predicting the actions of a clone of myself, it might be way easier to just query my own mental state than to simulate my clone. Is this similar to what's happening when LLM's are trained on LLM-generated data, as mentioned in the text?
> In my experience, larger models often become aware that they are a LLM generating text rather than predicting an existing...
(discussed on the LessWrong discord server)
There seems to be an implicit fundamental difference in many people's minds between an algorithm running a set of heuristics to maximize utility (a heuristic system?) and a particular decision theory (e.g. FDT). I think the better way to think about it is that decision theories categorize heuristic systems, usually classifying them by how they handle edge cases.
Let's suppose we have a non-embedded agent A in a computable environment, something like a very sophisticated video game, and A has to continually choose b...
SUMMARY: Get an AI within a sandbox to build another AI within the same sandbox. Then, figure out how it did that without getting eaten. I point out some problems with this approach.
Could we study successful inner alignment by observing it in action? My idea: Create a sandbox universe with sufficient complexity to allow for AI development. Place within it an AGI with some goal (we don't care what the goal is). If this AGI successfully builds and aligns a smarter ASI to achieve its goal without getting destroyed in the process, we can then query the AGI abo...
I love the xkcd book vibes of this post.
Incidentally, female chimps seem to live 25% longer than males—imagine human women lived until 90 while men died at 71.
Arithmetic error? both 71*1.25 and 71.5*1.25 to the nearest integer are 89, not 90. The error might (low-confidence, 10%) have been caused by calculating a 12.5% increase of 80 (exactly 90) and also dividing 80 by 1.125 (~71.1).
Why? Links appreciated.
There are some positive feedback loops in school that cause gaps in ability between students in a subject to widen. There are also some negative feedback loops (e.g., intervention), but the net effect is still the gap widening. Therefore, the system's behavior is chaotic (small differences in students' abilities eventually lead to big differences). If this is true, it means that some variation between students' successes is extremely difficult to predict.
Three examples of these positive feedback loops:
Suppose that Student A has less knowledge in a particul...
I would agree that the vibe is off.
It naturally follows that Eliezer Yudkowsky is so smart he can simulate himself and his environment.
Eliezer Yudkowsky never makes a rash decision; he thinks carefully about the consequences of every thought he has.
I think the definition of rational emotions as those based on correct thinking about reality is a bad definition--it makes both the act of modelling the world correctly and communicating your models to others slightly harder instead of slightly easier.
Imagine there is a faucet in front of me. Let's say this faucet is running hot water because I turned the right knob because I incorrectly thought the right knob turned cold water. It would be very strange to say that "The faucet's running of hot water is irrational."; no one would have a clue what you mean.
Y... (read more)