I’ve decided to try my hand on quantum mechanics sequence! Here’s what I have reached yet: https://app.suno.ai/playlist/81b44910-a9df-43ce-9160-b062e5b080f8/. (10 generated songs, 3 selected, unfortunately not the best quality)
ProgramCrafter
Karma: 88
This suggestion looks like it makes tradeoffs between different moral theories just like between agents. This is an usecase for ROSE values; they can be used instead of randomization to get even better results, where negative-fanatic-theories can’t virtually veto some actions: Threat-Resistant Bargaining Megapost: Introducing the ROSE Value.
It seems like a different definition of utility (“the sum of happiness minus suffering for all conscious beings”) than usual was introduced somewhere. Concept of utility doesn’t really restrict what it values; it includes such things as paperclip maximizers, for instance.
As well, agents can maximize not expected utility but minimal one over all cases, selecting guaranteed nice world over hell/heaven lottery.
Once the person already exists, it doesn’t matter what % of agents of a certain type exist. They exist—and as such, they have no reason to lose out on free value. Once you already exists, you don’t care about other agents in the reference class.
This means that you cannot credibly precommit to paying in a gamble (if coin comes up tails, you pay $1, otherwise you receive $20), since if coin comes up tails “you don’t care about other variants” and refuse to pay.
More than two hypotheses are already supported, you just need to close the “How to use” box.
Could you add probability logarithms (credibility/evidence decibels), please?
Another possible problem is that the different colored parts in the background would have to big enough and different coloured enough to store the information even after the screenshot is uploaded to different sites that use various compression algorithms for images.
Well, this is a problem for my approach.
Let’s estimate useful screen size as 1200x1080, 6 messages visible—that gives around 210K pixels per message. Then, according to [Remarks 1-18 on GPT](https://www.lesswrong.com/posts/7qSHKYRnqyrumEfbt/remarks-1-18-on-gpt-compressed), input state takes at least
log2(50257)*2048= 32K bits. If we use 16 distinct colors for background (I believe there is a way to make 16-color palette look nice) we get 4 bits of information per pixel, so we only have210K * 4 / 32K= 26-27 pixels for each chunk, which is rather small so after compression it wouldn’t be easy to restore original bits.So, probably OpenAI could encode hash of GPT’s input, and that would require much less data. Though, this would make it hard to prove that prompt matches the screenshot...
Probably message’s background could be made not plain color, but rather some hexagons or rectangles encoding conversation (most probably, tokens fitting into context window and output tokens). This way screenshots could be checked for being true. Also, this can provide way to detect data generated by ChatGPT and not train on it or move it to separate dataset.
The frozen neutrality
I’d say that it doesn’t carve reality at the same places as my understanding. I neither upvoted nor downvoted the post, but had to consciously remember that I have that option at all.
I think that language usage can be represented as vector, in basis of two modes:
“The Fiat”: words really have meanings, and goal of communication is to transmit information (including requests, promises, etc!),
“Non-Fiat”: you simply attempt to say a phrase that makes other people do something that furthers your goal. Like identifying with a social group (see Belief as Attire) or non-genuine promises.
(Note 1: if someone asked me what mode I commonly use, I would think. Think hard.)
(Note 2: I’ve found a whole tag about motivations which produce words—https://www.lesswrong.com/tag/simulacrum-levels! Had lost it for certain time before writing this comment.)
In life, I try to communicate less hyperboles and replace them with non-verbal signs, which do not carry implication of either “the most beautiful” or “more beautiful than everyone around”.
Continuing to make posts into songs! I believe I’m getting a bit better, mainly in rapid-lyrics-writing; would appreciate pointers how to improve further.
https://suno.com/song/ef734c80-bce6-4825-9906-fc226c1ea5b4 (based on post Don’t teach people how to reach the top of a hill)
https://suno.com/song/c5e21df5-4df7-4481-bbe3-d0b7c1227896 (based on post Effectively Handling Disagreements—Introducing a New Workshop)
Also, if someone is against me creating a musical form of your post, please say so! I don’t know beforehand which texts would seem easily convertible to me.
The LessWrong’s AI-generated album was surprisingly nice and, even more importantly, pointed out the song generator to me! (I’ve tried to find one a year ago and failed)
So I’ve decided to try my hand on quantum mechanics sequence. Here’s what I have reached yet: https://app.suno.ai/playlist/81b44910-a9df-43ce-9160-b062e5b080f8/. (10 generated songs, 3 selected, unfortunately not the best quality)
This post suggests a (and, quite possibly, the) way to select outcome in bargaining (fully deterministic multi-player game).
ROSE values replace Competition-Cooperation ones by having players not compete against each other in attempts to extract more utility but average their payoffs over possible initiative orders. A probable social consequence is noted, that people wouldn’t threaten everyone else in order to get something they want but would rather maximize own utility (improve their situation themselves).
ROSE values are resistant to unconditional threats. Unfortunately, conditional ones can’t be easily distinguished from conditional bonuses (if player B decides to pay A if A chooses some action from a certain set) so counteracting that requires further research. There are at least two more important directions of research: non-deterministic and imperfect-information games.
In comments, a problem is mentioned: what happens if there are multiple ROSE equilibrium points? I don’t believe solution “bargain over those points only” will work, as it’s possible all of those points will remain equilibriums when calculating ROSE values again. Hopefully, this players disagreement will involve only tiny slices of utility and not matter much overall.
Can’t Bob say “Alice is doing something strange; she is rejecting bet of smaller value at certain odds but accepting larger one, leaving free money on the table in former case, and that isn’t rational as well”?
Happiness is paradoxical. A 2011 study showed that people seeking happiness were generally unhappier than others.
This doesn’t seem paradoxical to me? Ones who are more happy than others are less likely to pursue additional happiness.
After all, if there’s a demon who pays a billion dollars to everyone who follows CDT or EDT then FDTists will lose out.
How does demon determine what DT a person follows?
If it’s determined by simulating person’s behavior in Newcomb-like problem, then once FDTist gets to know about that, he should two-box (since billion dollars from demon is more than million dollars from Omega).
If it’s determined by mind introspection, then FDTist will likely self-modify to believe to be CDTist, and checking actual DT becomes a problem like detecting AI deceptive alignment.
I guess that people who downvoted this would like to see more details why this “court” would work and how won’t it be sued when it misjudges (and the more cases there are, the higher probability of misjudge is).
(meta: I neither downvoted nor upvoted the proposal)
I’ve noticed that I’m no longer confused about anthropics, and a prediction-market based approach works.
Postulate. Anticipating (expecting) something is only relevant to decision making (for instance, expected utility calculation).
Expecting something can be represented by betting on a prediction market (with large enough liquidity so that it doesn’t move and contains no trade history).
If merging copies is considered, the sound probability to expect depends on merging algorithm. If it sums purchased shares across all copies, then the probability is influenced by splitting; if all copies except one are ignored, then not.
If copies are not merged, then what to anticipate depends on utility function.
“quantum suicide” aka rewriting arbitrary parts of utility function with zeroes is possible but don’t you really care about the person in unwanted scenario? Also, if AGI gets to know that, it can also run arbitrarily risky experiments...
Sleeping Beauty: if both trades go through in the case she is woken up twice, she should bet at probability 1⁄3. If not (for example, living the future: this opportunity will be presented to her only once), it’s coherent to bet at probability 1⁄2.
I’ve heard a comment that betting odds is something different from probability:
… what makes you think it [probability] should have a use? You can feel sure something will happen, or unsure about it, whether or not that has a use.
Well, if you feel sure about an event with incorrect probability, you may end up in suboptimal state with respect to instrumental rationality (since expected utility calculations will be flawed), so it’s perhaps more useful to have correct intuitions. (Eliezer may want to check this out and make fun of people with incorrect intuitions, by the way :-))
New problems are welcome!
Humans have never completed at large scale engineering task without at least one mistake
This line does not seem sufficiently well-defined.
Both “launch a satellite” and “launch the specific satellite” are large scale engineering tasks (or were, in the previous century); the first one had some mistakes, and in most cases the second one had no mistakes.
Transferring the argument to AI: the mistake may happen and be fixed while the task is not “create aligned AGI” but some prerequisite one, so it doesn’t ensure that the final AGI is unaligned.
When someone optimizes utility of a group of agents, all the utilities need to be combined. Taking sum of all utilities can create an issue where the world is optimized according to single agent’s utility function at the cost of others, if that function increases fast enough.
It’s probably better to maximize Ucombined=min(agent∈A)Uagent+arctan(max(agent∈A)Uagent−min(agent∈A)Uagent) - this way the top utility does not take over the full function since the difference can only add a finite amount of utility, but is still considered (so that there is incentive to improve the top utility if the minimal one has run into limitations).
Though, this creates a problem that agents would try to divide the utility function by some number to make their function considered more. It’s necessary to normalize the functions in some way, and I’m not sure how to do that.