Opinions expressed are my own and not endorsed by anyone.
Formerly @ ARC Evals aka METR
lukehmiles
Karma: 688
You may want to add MIRI’s botworld 1.0 project to the bibliography, so that people looking into this don’t duplicate the idea
Creating Environments to Design and Test Embedded Agents
Yes I agree it feels fishy. The problem with maximizing rubes is that the dilemmas might get lost in the detail of preventing rube hacking. Perhaps agents can “paint” existing money their own color, and money can only be painted once, and agents want to paint as much money as possible. Then the details remain in the env
Or something simpler would be that the agent’s money counter is in the environment but unmodifiable except by getting tokens, and the agent’s goal is to maximize this quantity. Feels kind of fake maybe because money gives the agent no power or intelligence, but it’s a valid object-in-the-world to have a preference over the state of.
Yet another option is to have the agent maximize energy tokens (which actions consume)
One way to test the “tasks don’t overlap” idea is to have two nets do two different tasks, but connect their internal layers. Then see how high the weights on those layers get. Like, is the internal processing done by Mario AI useful for Greek translation at all? If it is then backprop etc should discover that.
Zettelkasten in five seconds with no tooling
Have one big textfile with every thought you ever have. Number the thoughts and don’t make each thought too long. Reference thoughts with a pound (e.g. #456) for easy search.
Is the top one meant to be read first or last?
As with real and fake memories, I think if you’re careful then you can mainly deal with real ones
What about the info helper by itself? Is it much more useful in this context than it would be alone? Rewarding based on human prediction skill seems the best to me. I think Bostrom might have mentioned this problem (educating someone on a topic) somewhere.
I think crypto markets can’t be regulated except by random moderators’ filtering on bets and betters’ choices of where to put money. It seems someone could put a million dollars against a terrorist attack on a certain date and hope someone bets against it & executes to get the money. So a betting market allows hiring for certain tasks (not most tasks) with reliable verification & payout, and you get your money back if it doesn’t happen. I have some faith in moderators’ filters, though. I hope they would have the wisdom to forbid bets on terrorist attacks, assassinations, etc. Insider trading cannot be prevented (as far as I can tell) if betting is anonymous…
Printable PDF of whole sequence with comments
https://drive.google.com/file/d/1gW6btBWvk1mMCPItLt9wPYApc8caiUCa/view?usp=drivesdk
Against unstoppable crypto prediction markets
I think I can be a pretty good duck sometimes but where does that go on my resume? I need to get places with other skills but then i think my real move is duck
Imagine if you start with 100x stone and kinda quit any piece that doesn’t seem to be going pretty well when you half ass it and are not too picky about the final result. That’s a rough rough estimate.
You didn’t mean to but you’ve written a critique of using RCT’s to figure out medicine.
Mm i think “the higher process X is produced by SA” is different drill “the higher process X is implementing SA”
Definition paraphrasing attempt / question:
Can we say “a factorization B of a set S is a set of nontrivial partitions of S such that ” (cardinality not taken)? I.e.,
union(intersect(t in tuple) for tuple in cartesian_product(b in B)) = S. I.e., can we drop the intermediate requirement that each intersection has a unique single element, and only require the union of the intersections is equal to S?
That misses element 4 right?
>>> from itertools import product >>> B = [[{0, 1}, {2, 3, 4}], [{0, 2, 3}, {1, 3}]] >>> list(product(*B)) [({0, 1}, {0, 2, 3}), ({0, 1}, {1, 3}), ({2, 3, 4}, {0, 2, 3}), ({2, 3, 4}, {1, 3})] >>> [set.intersection(*tup) for tup in product(*B)] [{0}, {1}, {2, 3}, {3}] >>> set.union(*[set.intersection(*tup) for tup in product(*B)]) {0, 1, 2, 3}
We have but one metric of the real, price-independent, value of goods and services: Robin Hanson’s polls where he asks “would you own a home but die ten years early” and crap like that. You can compile his Twitter polls into some kind of constraints system and use it to fill in the true value of all things, extrapoterpoinferlating from other would-you-rather style polls. Then get Real Actual GDP
Thanks for pointing that out; I noticed that too. This is perhaps partly made up for by the fact that he doesn’t count feet or the mouth of the Nile river as possibilities.