@ryxcommar@andimgladofit I've never had more fun at any other time in my career. I can make refined, mature systems faster than ever. I'm branching out into more areas than I otherwise would. It's still not solved or too easy though. You can't build it all at once and care is important.
@TheStalwart There are very few pieces of software that you can put in a python loop that results in hacking access controlled systems.
This stuff is new and it's not something like they were just too lazy to apply standard practices here. The capabilities are novel!
@dbessner What is the delicious irony exactly? AI does best in software when driven by people who have learned how to code and exercise appropriate judgement.
@xwanyex Why don't you apply this logic to guns, white nationalists, neoconservatives, or Israel?
If the idea is that something might cause even the tiniest problem then we should close the door on it, then start applying that to other things.
@jaycaspiankang@PEWilliams_ Everyone is free to come to their own conclusions about how reputable a given opinion maker is based on their track record and expertise.
Is it antidemocratic for me to judge the current statements of Glenn Beck on the basis of things he's said in the past?
@zeanxe@BenjaminPDixon The only reason people are talking about this right now is because of the large support for solving a collective action problem and restraining frontier ai labs, both popular support and key people across institutions and the companies themselves.
@zeanxe@BenjaminPDixon Assuming that this is an inevitability and opposing legislation on that basis before it happens makes you look like a crank and also hands away any kind of influence that you might attempt to wield over this process.
@zeanxe@BenjaminPDixon If this were all a smokescreen for them to simply monopolize the business and crush mom and pop ai labs that would be a huge story if there were evidence of that. If that does come up in proposed legislation it will be a huge deal and people will make noise about it.
@zeanxe@BenjaminPDixon What is even the enforcement mechanisms you're imagining, how would these apply to open source LLMs?
It's insane to me that someone would oppose arms race style ai regulation on the basis of protecting mom and pop LLM shops. It's like farming all over again
@zeanxe@BenjaminPDixon 1. There's no proposal being put forward banning distillation. If there were that should be opposed.
2. Even if they did try to ban distillation, they can't stop china from doing distillation and these things have a way of getting out there.
@zeanxe@BenjaminPDixon What's the proposal to ban training methods? How could that even be defined in a way which wouldn't also hamstring the big companies?
@BenjaminPDixon If it's the case that the big players are writing the rules and the rules look like what Anthropic is announcing that they're self imposing, what regulation are you even worried about?
@BenjaminPDixon Ok, if you're an expert what is the specific policy that you think would entrench a monopoly and be an onerous burden on open weight models?