- Professionally, I do banking. It's a lot of integration work, sprinkled with a little algorithm every now and then. Lately I've been on capital requirements. The core of that is a system called AxiomSL, which is quite a lot of work for one guy to keep running.
In my spare time I write some algorithmic C, you can check that stuff out on github (https://github.com/DelusionalLogic) if you're curious.
I was an early adoter of LLM's. I used to lurk in the old EleutherAI discord and monitor their progress in reconstructing GPT-2 (I recall it being called GPT-J). I also played around a bunch with image generation. At this point nobody really tried applying them to code. We were just fascinated that it wrote back at all.
I have tried most of the modern models for development. I find then to generate a lot of nonsensical and unexplainable code. I've had no success (in the 30 or so times I've tried) at getting any of the models to debug or develop even small features. They usually get lost in some "best practice" and start looping on that forever. They're also constantly breaking style and violating module boundaries.
If i use them to generate documentation I find it to be surface level and repetitive. It'll make a lot of text about structures that are obvious to me just glancing at the code, but will (obviously) not have any context about the thought process that created that code, which is the only part I care about. I can read the code just fine myself. This is the same problem I find in commit messages generated with AI tools.
For the reversing I also do, I find the models to be too imprecise. It'll take large logical leaps that ruin understanding of the code I'm trying to understand. This is the only place I actually believe a properly trained (not a chatbot) model could actually succeed past the state of the art.
I don't really use stackoverflow either, I don't trust its accuracy, and it's easy to get cargo culted in software. I generally try to find my answers in official documentation, and if I can't get that I'll read the source code. If that's unavailable I'll take a guess, or reverse the thing If it's really important to me.
- > just ask any software engineer about what’s happening to their profession
I'm a professional developer, and nothing interesting is happening to the field. The people doing AI coding were already the weakest participants, and have not gained anything from it, except maybe optics.
The thing that's suffocating is the economics. The entire economy has turned its back on actual value in pursuit of silicon valley smoke.
- Transpiling shaders is what most game engines have done for a decade now. Everybody thinks it's stupid in that field as well, but there is no viable alternative.
- How would you identify "security researchers" and tell them apart from the attacker in a trench coat?
After you've done that, why would these supposedly expert security researchers review random code in your package manager?
- The exhausting "well actually" masks a corrosive argument, that if you can't enforce the rules in a rigid and rigorous fashion, the rule is fiat.
It's not that he doesn't know the difference. He's making the argument that since there's no _technical_ difference there can be no legal difference.
- Or maybe they wont and we will have ruined the one planet we can thrive to make a chatbot.
- Are we to assume that the people at the EFF haven't heard of how European nations, like Denmark, are building government infrastructure to verify your age without disclosing sensitive information?
Are we also at assume that the EFF fail to see the similarity of age-gating porn websites and age-gating entrance to strip clubs?
That doesn't seem likely to me, and I find it way more likely that the EFF is purposefully excluding the best argument against their chosen position.
- We have newer and more relevant data than DARE.
- My guy, this is making the opposite argument from what you think:
"On the illegal market, no one is checking IDs before selling marijuana. When and where cannabis is illegal, high schoolers often sell cannabis to their peers. In contrast, licensed cannabis stores have overwhelming compliance with age-gating."
It has indeed not increased the cannabis use of kids, but that would also still be illegal. That study is an argument that age gating works.
- > How many kids say “I really want to smoke weed but it’s illegally so I won’t do it”?
I think it's generally accepted that marijuana use increases after legalization. So yes.
- A lot of the proposals don't involve you sending your drivers license or "other information" to anyone. The site in question asks you to verify with a trusted third party (usually a government entity), and that trusted third party only provides then with the end result of the validation.
> which is what Google etc have been trying to do for years but this would just completely fast track that.
Excuse me? They have done that for years. There's nothing to "fast track" here. Big Tech already implemented surveillance.
- > I can absolutely guarantee you that any teenager can easily get access to weed, cigarettes and alcohol
Is you argument then that we shouldn't age gate those things in reality either? Would you suggest that teenagers smoke and drink just as much as they would have had it been legal to sell to minors?
Laws don't just exist to stop you, they also exist to shape society. They exist as signals for what we deem appropriate behavior.
- Cool trick to tie in the libertarian idea of protecting yourself from legally sanctioned government actions.
- The states (or rather the national banks of said states) are usually the ones running the central clearing system. That's the place where all the different banks report their net change in relation to all the other banks, and settle that change on their account with the central bank.
Believe it or not, banks don't ferry around cash to each other. It's all just numbers in a computer.
- In my opinion, this isn't a problem of AI. the people who get deceived by this are willing participants in the lie. When proven wrong, they will fall back to the echo chamber and rely on it to give them more false facts. They won't seek information outside of their own circle. They cannot be understood as merely passively misinformed. They are actively lying to themselves.
- What you've identified here is a core part of what the banking sector calls the "risk based approach". Risk in that case is defined as the product of the chance of something happening and the impact of it happening. With this understanding we can make the same argument you're making, a little more clearly.
Cloudflare is really good at what they do, they employ good engineering talent, and they understand the problem. That lowers the chance of anything bad happening. On the other hand, they achieve that by unifying the infrastructure for a large part of the internet, raising the impact.
The website operator herself might be worse at implementing and maintaining the system, which would raise the chance of an outage. Conversely, it would also only affect her website, lowering the impact.
I don't think there's anything to dispute in that description. The discussion then is if cloudflares good engineering lowers the chance of an outage happening more than it raises the impact. In other words, the things we can disagree about is the scaling factors, the core of the argument seems reasonable to me.
- > If I were an asshole, I would have trained them poorly and drug it out.
That's of course the obvious way this goes wrong. Bad intentions. The much more insidious version is that you could have just been a terrible teachers, maybe you suck at training your peers, and you don't know.
The end result is the same. You look like the only person who gets it amongst the riff-raff, but in this case you don't even have a choice. The system has produced a poor outcome not because anybody abused it, but because it was a bad system.
- I do this because I'm a non-native english speaker. My preference varies from word to word. I write color, but i also write aliminium.
Those are just not realistic numbers.