hereisonehand
Did you all see this? https://twitter.com/SquishChaos/status/1383435339910418432?s=20
Basically, claiming in the next 12 months ethereum will undergo the supply shock equivalent of 3 bitcoin halving events. Curious if rationalists see a flaw with the reasoning or are already ahead of this
I have been watching this video https://www.youtube.com/watch?v=EUjc1WuyPT8 on AI alignment (something I’m very behind on, my apologies) and it occurred to me that one aspect of the problem is finding a concrete formalized solution to Goodhart’s law-styled problems? Like Yudkowsky was talking about ways that an AGI optimized towards making smiles could go wrong (namely, the AGI could find smarter and smarter ways to effectively give everyone heroin to quickly create lasting smiles) - and it seems like one aspect of this problem is that the metric “smiles” is a measurement for this ambiguous target “wellbeing,” and so when the AGI gives us heroin to make us smile we go “well no, that isn’t what we meant when we said wellbeing.” So we’re trying to find a way to formally write an algorithm for pursuing what we actually mean by wellbeing in a lasting and durable way rather than an algorithm that gets caught optimizing metric that previously measured wellbeing before they were optimized so much. I get that the problem of AI alignment has more facets than just that, but it seems like finding an effective way to tell an AI what wellbeing is rather than telling it things that are currently metrics of wellbeing usually (like smiles) is one facet.
Is this in fact a part of the AI alignment problem, and if so is anyone trying to solve this facet of the problem and where might I go to read more about that? I’ve been sort of interested in meta-ethics for a while, and solving this facet of the problem seems remarkably related to solving important problems in meta-ethics.
Thanks for the feedback. Will change the format in the future!
Is this sequence going to become a sequence in the lesswrong content library at some point? I kind of like having things in the library page so I can go back and read the whole thing later but I noticed it’s not there yet.
woot woot! that long-term thinking paying off!
Is there a deadline? I’m a bit busy until the end of this week, but I wanted to try my hand at doing some lit review soon anyways. I’ll probably take a shot at this sometime next monday.
Another weird takeaway is the timeline. I think my intuition whenever I hear about a good idea currently happening is that because it’s happening right now, it’s probably too late for me to get in on it at all because everyone already knows about it. I think that intuition is overweighted. If there’s a spectrum from ideas being fully saturated to completely empty of people working on them, when good ideas break in the news they are probably closer to the latter than I give them credit for being. At least, I need to update in that direction.
The fact that Amazon rainforest produces 20% atmospheric oxygen (I read this somewhere, hope this isn’t fiction) should be a bigger political piece than it seems to be. Seems like brazil could be leveraging this further on a global stage (have other countries subsidize the cost of maintaining the rainforests, preventing deforestation as we all benefit from/need the CO2 to oxygen conversion)? Also, would other countries have a tree-planting supply race to eliminate dependence on such a large source of oxygen from any one agent?
Just a strange thought that occurred to me this morning. Obviously doesn’t reflect unfortunate realities of current politics (it isn’t much of a political piece if the other agents don’t believe it’s real), but it occurred to me as a alternative politics that might transpire in a world where everyone took climate change seriously as an existential threat.
Have you checked out any work on coherentist theories of epistemic justification? I definitely haven’t done the work to have an opinion on this, but I remember this dichotomy (foundationalism v. coherentism) being referred to in old introductory epistemology coursework.
cool. I’ll be sure to get something in by the 16th.
Ya, it’s interesting because it was a “so clearly a good idea” idea. We tend to either dismiss ideas as bad because we found the fatal flaw or think “this idea is so flawless it must’ve been the lowest hanging fruit and thus have already been picked.”
Another example that comes to mind is checklists in surgery. Gawande wrote the book “checklist manifesto” with his findings that a simple checklist dramatically improved surgical outcomes back in 2009. I wonder if the “maybe we should try to make some kind of checklist-ish modification to how we approach everything else in medicine” thought needs similar action.
Yup, it was a quick thought I put to page and I will quickly and easily concede that 1) my initial idea wasn’t expressed very clearly, 2) the way it was expressed is best interpreted by a reader in a way that makes it non-sensicle (“what does it mean to say oxygen is produced” and I didn’t really tie my initial writing to climate change in the way I wanted too so what am I even talking about), 3) even the way I clarified my idea later mixed some thoughts that really should be separated out (viable != effective), and 4) I have some learning to do in the area of EA mental models and reasoning about public interventions. Not my best work.
Reflection:
I’m messing around with shortform as a way to kinda throw ideas on a page. This idea didn’t work out too well towards generating productive discussion as upon reflection, the idea wasn’t super coherent, let alone pointing towards anything true. However, I got a lot more engagement than I expected, which points to something of value from the medium. I think the course forward is probably to 1) keep experimenting with short form because I gain something from having my incoherence pointed out to me and there’s a chance I will be more coherent and useful in the future, and 2) maybe take 5 minutes to reread my shortform posts before I post them (just because it’s short form doesn’t mean it can be nonsense)
Thanks for helping me get informed. I was under the impression that (and this is a separate thread) planting trees was a viable initiative to fight climate change, and by extension the survival of the amazon rainforest was a significant climate change initiative? I guess I’m wondering if along these lines as well, if climate change is important on the world stage, then the health of the rainforest would be as well?
**Thanks for correcting me about the oxygen consumption line—that is what I said and it was misguided
From your post:
This is circular, but is this necessarily a problem? If your choice is a circular justification or eventually hitting a level with no justification, then the circular justification suddenly starts looking pretty attractive.
I think the coherentist article I linked to has some useful perspective here. The quote below is from a section in that article on regress. The first paragraph outlines a view similar to yours and raises an important objection against the circular justification view. The 2nd paragraph raises a potential response.
What is the coherentist’s response to the regress? The coherentist can be understood as proposing that nothing prevents the regress from proceeding in a circle. Thus, A can be a reason for B which is a reason for C which is a reason for A. If this is acceptable, then what we have is a chain of reasons that is never-ending but which does not involve an infinite number of beliefs. It is never-ending in the sense that for each belief in the chain there is a reason for that belief also in the chain. Yet there is an immediate problem with this response due to the fact that justificatory circles are usually thought to be vicious ones. If someone claims C and is asked why she believes it, she may reply that her reason is B. If asked why she believes B, she may assert A. But if prompted to justify her belief in A, she is not allowed to refer back to CC which in the present justificatory context is still in doubt. If she did justify A in terms of C nonetheless, her move would lack any justificatory force whatsoever.
The coherentist may respond by denying that she ever intended to suggest that circular reasoning is a legitimate dialectical strategy. What she objects to is rather the assumption that justification should at all proceed in a linear fashion whereby reasons are given for reasons, and so on. This assumption of linearity presupposes that what is, in a primary sense, justified are individual beliefs. This, says the coherentist, is simply wrong: it is not individual beliefs that are primarily justified, but entire belief systems. Particular beliefs can also be justified but only in a secondary or derived sense, if they form part of a justified belief system. This is a coherence approach because what makes a belief system justified, on this view, is precisely its coherence. A belief system is justified if it is coherent to a sufficiently high degree.
Of course, the key coherentist claim is that an entire belief system can be justified without individual beliefs being justified because the property of being justified is a property of belief systems that emerges from the coherence of multiple beliefs.
I keep seeing these articles about the introduction of artificial intelligence/data science to football and basketball strategy. What’s crazy to me is that it’s happening now instead of much much earlier. The book Moneyball was published in 2003 (the movie in 2011) spreading the story of how use of statistics changed the game when it came to every aspect of managing a baseball team. After reading it, I and many others thought to ourselves “this would be cool to do in other sports”—using data would be interesting in every area of every sport (drafting, play calling, better coaching, clock management, etc). But I guess I assumed—if I thought of it, why wouldn’t other people?
It’s kind of a wild example of the idea that “if something works a little, you should do more of it and see if it works a lot, and keep doing that until you see evidence that it’s running out of incremental benefit.” My assumption that the “Moneyball” space was saturated back in 2011 was completely off given that in the time between 2011 and now, one could have trained themselves from scratch in the relevant data science methods and pushed for such jobs (my intuition is that 8 years of training could get you there). So, it’s not even a “right place, right time” story given the timeline. It’s just—when you saw the obvious trend, did you assume that everyone else was already thinking about it, or did you jump in yourself?