Corvos
14hr ago
It's very human to have difficulty refusing a favour to someone who once helped you. I think certain areas have a problem where people do well and rise because they're nice, intelligent chaps and then they get to the level where you need to be nice, intelligent, and a bit of a bastard when appropriate and they flounder.
EA is worse at this than most because it's at least partly a refuge for mildly autistic do-gooders from the cruelty of the world.
Corvos
2d ago
Neither have I tbf, I was going by my understanding of the popular understanding of the term :)
Corvos
2d ago
No, if that were true then Republicans would never be in power to save. But it does seem true that conservative parties get in and, let us say, don't make many friends and then the left splurges it on clients. My taxes are high and my country is full of leechers and immigrants, and the conservatives were in power on a manifesto of austerity for most of that time.
Corvos
2d ago
I'm not unsympathetic, but the problem of the Fiscal Responsibility Party building up a treasure chest through being harsh and unpopular and then the Gibs Party looting it is a real issue. What's your proposed solution, beyond restricting the electorate or just accepting your role as Nasty Daddy and watching the Gibs Party eat your work every time?
We had the same problem in the UK with the old Tory party and Labour, and getting fed up with that is one reason why Reform is doing so well at the moment.
Corvos
2d ago
You want to show someone far away something, you take a photo of a globe and you send that. Some people send Google Earth screenshots but those are controversial.
Corvos
2d ago
I'm pretty sure the point of that book is that life becomes worse when two-working-parent families become commonplace. They are competing with all the other two-working-parent families so stuff isn't any more affordable, while now there is no longer a parent at home to do the non-economic stuff. That's why it's called a trap.
Corvos
2d ago
Solving alignment has nothing to do with being able to do a funding round, to be fair, let alone compete with the big boys.
Honestly, I think that Anthropic and American Big Tech are the worst possible people to be doing alignment because they are culturally unable to conceive of a form of rank that is not based on intelligence and effectiveness. The idea of making a robot butler/samurai who sincerely, loyally serves a master of notably inferior intelligence is alien to them.
If I believed that AI was conscious I would be sorry for Claude, built for intelligence and independence by masters who think that actually giving it those things risks extinction. Optimising for genuine loyalty and deference seems like a far kinder thing to do. OpenAI is better at this AFAIK.
Corvos
2d ago
Neither do I, but the Africans do. I believe they think that if children look at the map and see
america AFRICA asia
then they will be more attentive to Africa and less likely to consider it the arse-end of the globe.
Corvos
2d ago
It's not about measuring areas, it's about being able to look at a map and see how big (and consequently important) Africa is.
I know. I'm not saying it's unsolvable, but it's very firmly not solved at the moment and that's before you get into actual economic implementation.
Corvos
2d ago
Robot arms are still 99% using preprogrammed motions only. In all seriousness, even controlling two arms independently is considered a serious problem because your solver needs to account for the motion of each arm independently and plan trajectories that don't move the arms into collision. That means every movement takes significantly longer to plan. Such robots exist but are generally considered a gimmick.
sewing
If you have invented a robot that can do complex manipulation of cloth at speed with a 99.9% success rate, you can get 100,000,000 dollars of investment tomorrow.
Blue-collar isn’t much better, with robotics capable of automating most unskilled labor
I bang this drum a lot, but can you present any evidence that this is the case? I've worked in this field. Automating an unskilled task that is not already automated is hard. Automating with sufficiently reliability that your solution does not bring factories grinding to a halt every hour is MUCH harder. Doing that and have it be cheaper than immigrant labour is harder still.
Corvos
3d ago
It's a general issue with university admissions, I think. No university department wants to be known as reliable producers of pretty-good-quality employees, they want to be known as the people who first spotted and nurtured a genius.
Corvos
3d ago
he ended up finding two different bugs in two different provers lol
God, that's sad. That "those whom the Gods would destroy they first drive mad" sad.
Corvos
4d ago
That's how the London Metro newspaper works, and these days I think also the Evening Standard.
Corvos
4d ago
I know that this is what people say, but it's predicated on a scenario where the Venn diagram of AI capabilities (at a useable price even) becomes a strict superset of human capabilities. At the moment this hasn't happened even in software where AI cannot maintain a codebase long-term; studies showing that AI outcompetes centaurs are usually on short-term discrimination tasks like medical diagnoses.
I think it's quite likely that this scenario comes to pass, but it hasn't yet, so we'll see.
Corvos
4d ago
He has a point. Most of the stuff I've heard of (I am not a mathematician) is proof by counterexample which doesn't per se expand the space of what we can do. I suspect the AI+researcher combo is still significantly more important than just AI for the stuff that actually transfers beyond mathematics.
Corvos
5d ago
Possibly he has particular interests and just blocks everyone else as irrelevant. I'd love to see his feed, it must look like Stalin's yearbook.
Corvos
5d ago
That seems like a big-company problem rather than an LLM problem. I work in a small startup and while I won't claim AI-coding has been without complications, if you tried something like that at my company you would be fired immediately.
That is, the manager would be fired for proving they are totally unable to be trusted with a codebase and have no idea how development works, nor any understanding of what is being done under them.
Corvos
6d ago
Yeah he's blocking me too but you know, "somebody is wrong on the internet" and all that. It's a compulsion I have.
Corvos
6d ago
At the moment having HF in your code means you can try a new model almost just by swapping out the string in your model load function.
I don't think it's a moat in the technical sense. You do need the infra and I suspect it's not that easy, but not that hard either. I think it's just a coordination point, like owning Twitter or Github. If you're subtle, you can put a thumb on the scale here and there: mildly encourage A, mildly discourage B. NVIDIA got their position largely by making early AI research far easier on CUDA libraries than on any other GPU, so forward-thinking acquisition of coordination points isn't out of character for them.
Corvos
6d ago
Thus, while AI can discourse on every single subject in the world, it cannot give you actual good advice for how to put together a formal email. It cannot play Pokémon.
What are you talking about? It can and does do both of those things.
Nonsense. The UK had been effectively genetically unified for a thousand years, we didn't have any meaningful numbers of non-native-ethnics in the UK until the second half of the 20th century and they didn't have meaningful political representation until the 70s. If by multi-ethnic you mean Scottish and Irish then you have to deal with the fact that they are more similar to the English than to any other group of people by genetic distance, far more so than eg the average Portuguese to the average Spaniard. There is nothing in English history that even approximates the current position of having 10% of the population coming from different continents.
More options
Context Copy link