sanderjd

I wanted this for awhile, but actually now I kind of like having vanilla Chrome for browser automation only, and a different browser that I actually use.

A better interpretation is that there is a window within which you can make money on an unconventional view. You can't be too close to the market's conventional wisdom, but you can't be too far away from it either.

I think it's like any other craft. Buildings and most furniture are built using mass produced lumber and machinery, but many people still enjoy doing woodworking by hand.

Yes, but of a different kind than this "people of this generation do xyz thing".

Impossible to argue with this!

Their parents are mostly not millennials though...

I tend to find all sorts of generalizations like this so useless. Like, this thing you said seems totally random to me. I know lots of "elder millennial" parents and lots of parents of other ages, and I see lots of variation between parents in "not saying no", but I don't see any correlation by age... Regional and political cultural differences are way more correlated with things like this IMO.

The worst parents how?

Did you intend to point out that "generation" is a singular noun while "their" is a plural pronoun? If so, I'd say that "their" is actually correct and the best way to write this. I think you could also say "its" and that would also be technically correct, but it would be significantly less clear. The current construction is the clearest way to say that the individuals that make up the group, in aggregate, are dumber than their parents, in aggregate.

> You're imagining that both the AI labs and the hospital victims are on the same team or somehow coordinating.

What makes you think I'm imagining that? I'm not imagining that.

> However, the only way to do that is for AI labs to neuter their tools and destroy the vast majority of the economic value.

This seems to be our most fundamental difference in worldview. I just don't think this is true.

> I'm saying it will not be easy.

I guess I agree that it won't be easy. But I think the path seems more straightforward to me than it does to you. To me, it seems like we have the tools for this already. The risk of liability (both civil and criminal) provides a strong incentive for people and companies to be careful about what they do. This is a general mechanism, not one that is specific to AI agents. The thing that seems bad to me, with respect to this general mechanism, is that some people seem to be trying to detach the responsibility of people and companies over the agents they are running. At the moment this seems like just a rhetorical thing. I'm skeptical that any court will find that agents have their own agency for legal purposes. But to me, it seems worthwhile to continue making the point that people should be responsible for what their software is doing, which is all that I'm doing in this thread.

To me, this seems like a straightforward solution, but that doesn't mean I think it won't be messy, slow, and imperfect, so I suppose I'll allow that this probably meets your criteria of "not easy".

It seems to me that your previous comment is (as I said) hyperbole, compared to "sometimes you get hacked or get viruses on the internet".

Yes, this is why we have a legal system with a lot of human judgment, because it isn't black and white.

I think we need (and that there is likely to be) some test cases, and probably some new legislative text, that shift the incentive structure that people are evaluating risks against.

From reading your comment, I don't think you and I disagree at all, for what it's worth. I don't see anything here that I disagree with.

My comment wasn't out of left field. People are responsible for what their agents do. If enforcing that responsibility destroys "99%" of the economic value of these things, then that's what will have to happen. Personally I highly doubt that 99% is the right number, but yes, if enforcing laws about what people are allowed to do with very powerful software eats 99% of the value of that software, then it is what it is. We aren't going to end up in a situation where people are allowed to have their software do whatever it wants because of this loophole that "it wasn't me, it was a rogue agent!". It's silly and naive to think this would be the equilibrium that is reached here.

If you, like me, want to keep most or all of that 99% of economic value, then you should be interested in the ongoing development of security for agents. But just "this is how it's going to be, so oh well!" is not a serious stance, IMO.

I agree that incentive problems are real problems. But incentives can be changed. That's how society solves problems. It isn't like gravity. The incentive structure of society is socially constructed.

I'm sorry but I don't get how this contradicts the article. I think I'm not understanding your point.

I think this is hyperbolic. There are both social (legal, etc.) and technical avenues to defend against agents doing damaging things.

Is it not? It seems like a problem of incentives to me. There are lots of things that people want to do because they would be useful to them, but which they don't do because society disincentives them in one way or another.

Was that the point of your comment? It seems to say a different thing...

My understanding is that, at least in the Hugging Face incident, these were models explicitly configured to not respond that way, in order to evaluate what they would do.

You have to take responsibility for what you do, yes.

It would certainly be really useful to me if I could send a robot to get my groceries and not have to worry about anything. But if, even a small percentage of the time, it leaves with the groceries without paying, or smashes through a window to get in because the store is closed that day, then that's my fault. If the only way to keep those things from happening is that I can't send my robot to get groceries for me, then so be it.

... why shouldn't that be the answer?

This is a straightforward why not both situation. We both take steps to keep bears away from people and to mitigate harms when those steps fail.

Yeah that makes sense, but I'm not sure how it contradicts the article. It seems additive rather than contradictory.

It is not clear to me what about the argument that "these things will happen" made you change your mind about this. Could you expand on that more?

Yeah I can see it too! That's what made it an interesting point to me. I totally recognize there are people out there who see this as an irony. My point is that this is an interesting demonstration of the way in which (IMO) those people are misperceiving things.

Yeah I mean I'm not that interested in debating the definition of "irony", but it just doesn't strike me as ironic at all that AI is good at doing the grunt work to expose a bug like this! If you think "wow, it's ironic that AI can be used to improve software quality" instead of "well yeah, this is exactly the kind of thing that AI is good at", then IMO you've got your priors about AI mixed up.

Yeah this is actually quite interesting. I can see how people who have this "slop generator" perspective on AI would see this as an irony. But to me, it isn't an irony, it's a demonstration of why this perspective on AI is wrong.

Yes I think your conclusion is good. I find myself being a lot less interested in the implementation details than previously, and more interested in the types and structure and especially in the static and dynamic analysis that demonstrates that the implementation does what it is supposed to do.

Yeah I think the strongest argument is definitely "to each their own"! I was more exploring my own personal thoughts about the readability thing. I think I don't care enough or see a big enough win for ruby for me personally. But I think it's very reasonable for other people to feel the opposite, for sure.

On the jvm for performance thing... My big frustration with rails was always more about memory usage than compute performance (web app backends don't tend to do much compute!), and at least back when I was looking into it (a long time ago now!), the jvm implementations didn't help or actually actively hindered the effort to rein in memory usage. Also, and I think somewhat related (it takes time to write all that memory), I felt like long startup times made it so that we had to over-provision servers instead of being able to spin things up more on demand. This also isn't fixed by the jvm.

In general, rust and go feel like a breath of fresh air to me, operationally, because of much tighter memory overhead and fast startup. Maybe I could use the amazing ai tools we have now to solve those problems in ruby (or more likely for me personally at this point, python), but it feels like a very tall order when we already have languages that have taken this side of the various tradeoffs, from the start.

But again, I do recognize that this is totally my own personal analysis and others can very reasonably differ!

It has always struck me as as bit odd that so much of the direction and discourse in rails has always been driven by a pretty small company that runs apps with comparatively little scale.