PostHog has this feature today. They have the concept of "Scouts", which are agents that scour ingested event data, replay sessions, and possibly also other kinds of data (though I haven't looked into what that means in practice), triage what they see, and put up PRs against them.
Update: I hope that didn't come across as a detraction; it's cool that you're building this. PostHog is open source, so if you haven't looked at their implementation yet, take a look, maybe it'll be useful to you!
Not sure if this is productizable as-is, but I think it's absolutely the right direction. A lot of software has to become WAY more malleable directly in the hands of users.
Software vendors should be providing the platform, including any hard-to-build widgets, and "stock" look-and-feel so that the software is usable out of the box, but then the customers should be able to send their product feature requests directly to a chatbot, whether it is embedded into the platform (i.e. same vendor) or rides on top (i.e. different vendor).
This is the new form of UGC. The vendor can turn it into a marketplace within that platform, promote popular 3rd party features in that marketplace, etc.
When high quality effort is applied to a tool, such as AES or the linux kernel, we intuit that it "hardens" the tool. That is, it makes the tool more correct, more resilient, less assailable, etc.
Similarly, when effort is applied to an open problem, such as the Riemann hypothesis or P v NP, without progress, it "hardens" the problem: it makes the problem feel more daunting to whoever takes a stab at it next.
Andrew Wiles, whose interview also hit the homepage today (https://news.ycombinator.com/item?id=49075264), couldn't just tackle Fermat's Last Theorem head on, he had to wait until a different, modern problem reduced to it, because FLT had gathered this mystique of unassailability through its 300 years of existence.
A thing I worry about is that as AI transmutes tokens into effort, it'll split the world into two: some problems will yield, making human effort entirely unnecessary, and others will harden to the point where human effort will feel increasingly less worthwhile, because "even AI couldn't solve it". I don't like this. AI is spiky, so I suspect it'll continue having major blind spots, and yet its mere presence will probably have a chilling effect on what would have otherwise been useful human effort.
1. Some of the "AI" proofs applied existing human work from lesser-known papers. AI proofs could solve the long-standing problem in math of almost all attention concentrating on less than 1% of authors. Human effort from the other 99% would have otherwise been wasted, which AI can rescue and give credit to thanks to its superhuman ability to match patterns across reams of text.
Humans may remain superior in spatial / non-verbal reasoning for a while longer yet, and, in the meanwhile, computers may aid us in collaborating to put that to use better.
2. AI-assisted, computer-verified proofs could further democratize mathematics by reducing the power of connections to get a reviewer to look at a journal submission. We can then also decouple the two tasks of
a. Verifying a statement is true
b. Explaining it
3. Searching for previous work and finding the edges of human knowledge are now easier. And we can leap across tedious terrain that the machine has the patience to plod through to find more interesting questions.
This is a problem that will solve itself, people will continue to work on the problems that AI fails at, likely by telling AI the approaches they want AI to take.
AI is nowhere near the intelligence of a very educated person that has innate talent for problem solving. It does solve the problem of applying human intelligence on problems that truly need it. AI is also a great tool to see if there's something simple that we've missed or just haven't even attempted due to wrong assumptions.
In this case it's not actually that big a result, it's an incremental improvement on a series of previous results. Cryptographer Orr Dunkelman describes it better than I ever could, he's one of the people who produced one of the previous results:
The main result in this paper is improving the Derbez, Foque, Jean attack from EUROCRYPT 2013, which is an improvement of our attack from CRYPTO 2010, which is an improvement of the Demirci-Selcuk attack, which is the improvement of the Gilbert-Minier collision attack against 7-round attack [...]
To save everybody's time, the [DFJ13] attack is on 7-round AES. The new result is also on 7-round AES, "eroding" the security margin of 7-round AES by about 8 bits of security [...] While this is the first improvement in attacking 7-round AES in the last decade, if you were not worried by the series of papers that reduced the security of 5-round AES from 2^32 to 2^16, or the somewhat improved attacks on 6-round AES, then you should not really worry now to start a procedure for changing 10-round AES (for 128-bit key) for something else, when there are no attacks on 8-round AES-128.
So someone threw a clanker at a series of previous results and told it to find improvements. Since it's ingested every piece of crypto research ever and can draw on all of them instantly, it managed to tweak the previous work a bit to improve the attack slightly... on a version of AES deliberately weakened to make attacks easier, a standard procedure for any iterated crypto algorithm where you see how many rounds you can get into it before your attack stalls. So it's a bit like saying you knocked out Mike Tyson's brother's cousin's uncle's sister's nephew in four rounds instead of five.
i think id almost worry more that ai can solve problems in latent space that it cant translate back to tokens because decoding ruins it, and that we wont be able to come up with concepts that we can map to properly decode those solutions in a way people understand
... what is understanding of mathematics anyway? if some AI result helps a mathematician to solve more problems I would say then that it gave them some understanding, but just as there are proofs that span hundreds of pages it's likely that soon proofs will be long Lean programs and studying them will be part of mathematics, just as studying Go played by AI.
This blog post talks in depth about what you're talking about. It may interest you. It even talks about the future where math proofs are just Lean programs, and why that won't necessarily be a good thing.
I have complicated feelings regarding academia, and in general IMHO it's way past due to start focusing on quality instead of hype.
who was first to some kind of novelty? who cares. someone did something but no one can replicate it? intentionally wasting public money. fraud by any other name.
"science" wouldn't move slower if we would build more robust data generating processes.
of course, since usually it's hard to judge quality academia uses proxies. not to mention that the people who could usually are also live inside fancy glassware. and it would be a shame to rock the boat.
... but math is doubly special, because we accepted that it doesn't matter (until it does, but then it's called cryptography and logistics network optimization and high frequency trading, and machine learning), and how long a problem stays unsolved was quite a good proxy.
still, if AI solves the easy ones we can finally have fun with the hard ones!
They’re likely referring to his post / thoughts on the bun rust rewrite and specifically those he shared about the lead developer of the bun project and the bun project at large.
I would not characterize what he shared as vitriolic, but from the comments when the post was submitted one would get the impression that is not a majority opinion here.
The comments here indicate a surprisingly, in my view, mixed reception, for such an unambiguously good talk?
The main thrust is simple:
- Technology is not good or bad, it is an amplifier of human values and our collective ambitions.
- Those of us who are terrified of where technology is going are actually terrified that humans are, on the net, bad, and technology will amplify that, bringing about dystopia.
- Believing this this is self-fulfilling. By embracing such beliefs you surrender your agency, which is what actually brings about dystopia. If you, instead, have a positive vision of humanity, then you are one of the people that drives society forwards rather than backwards.
We are currently at a moment in history where this message bears repeating.
A lot of things feel unstable, and as a result we have an opportunity to redefine the world in good ways and bad ones. Those who choose to redefine it in ways that are intractably bad will often do so from the point of view of a system observer, not a system participant. This is cope, because in reality everyone, whether they choose to accept it or not, is a system participant.
> The comments here indicate a surprisingly, in my view, mixed reception, for such an unambiguously good talk?
It wasn't an unambiguously good talk. Many of the people attacking various portions of the talk are making good points. I actually agree that blackpilling is a self-fulfilling belief and indulging in it weakens your own agency, and it's better to have a positive vision of humanity and to wield your own agency towards that end. But the thing about positive visions of humanity is that different people have different, mutually-incompatible positive visions; and I think a lot of people don't necessarily agree with Andrew Kelly's specific one.
I don't really think the main thrust is correct, though, if that's what it really is. Because humans don't have to be bad for tech to bring about a dystopia even under your first premise. They just have to have enough instincts that are even potentially adaptive in times of scarcity that become maladaptive in a high-tech environment for tech to bring about a dystopia. And I don't think that dystopia will be a self-fulfilling prophecy, even if your second premise is correct, because if humans have a sufficient number of instincts to interact with technology to bring about a dystopia, it's very likely to happen whether we believe it or not.
Finally, even the majority can be optimistic about humanity or even everyone, and it could still lead to a tech dystopia because of the additive effects of good intentions sometimes adding to bad (see fossil fuel use for example).
I don't think we have many opportunities to redefine the world. We're already in a state where people just trying to get by is contributing to the destruction.
> - Those of us who are terrified of where technology is going are actually terrified that humans are, on the net, bad, and technology will amplify that, bringing about dystopia.
I think this is true in that people will do what they need to in order to get ahead. If you play a game by the rules against someone who is cheating, then you will lose. The only way to stay competitive and have a chance to win is to also cheat. When people see others who act in an anti-social way (PE destroying companies, enshittification of software, etc.) go unpunished and reap massive rewards...then they too have to be anti-social just to stay on the same level or have a chance of succeeding. It's the descent into a low-trust society and a general race to the bottom.
The hope for me was that humans have learned from history, from all the past mistakes of failed civilizations and collapse of authoritarian regimes to build a modern world that is equitable. But there will always be narcissistic psychopaths who are able to say the right words and sweet talk those they need to in order to gain power and repeat history.
You're surprised they didn't eat the tokens to churn on lots of open problems instead of asking others to pay for those tokens? They're in the token business. If they're eating the tokens, it's in support of a marketing effort, not in support of innovation across the frontier of all the other academic disciplines. The collective frontier is way too big for them to just "solve it" without asking society to at least help them break even on such an enormous public good.
It's also much better to distribute the challenge of identifying problems amenable to which prompts
Could they do it? Sure, but to what end? It would make more people hate them and feel even more "take our interesting work." Pitching it as a useful tool just makes more sense on all levels
reply