Look, I get there are actual risks to consider with distillation attacks
But the frontier labs complaining about it will always be really, really funny
We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax.
These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.
I actually use AI the perfect amount. Anyone who uses it less than me is a hopeless Luddite destined for the underclass, anyone who uses it more than me is mindlessly degrading their literacy and erudition.
@stephenjhall And frankly, the even more pessimistic take is that these "fast" versions of models are not dropping to help SWEs, they are dropping to part vibe coders with their money quicker.
My brain is so broken from the internet that when I see weird takes like this, I can never tell if it's like super dry irony or their actual take and its just weird.
LOL.
I’m really sorry guys, but the AI bro “something is happening” influencers are correct.
If you don’t agree, you don’t know how hard it is to make a program that just makes other programs with 5 words of English in 5 seconds.
This is moon landing stuff.
@stephenjhall Anyways, actual response is that I don't think this is "moon landing stuff", for the same reason a lot of SWEs have been shouting from the rooftops the past few months, LoC isn't the primary bottleneck in work. It's certainly cool and and improvement, but not game changing.
@stephenjhall Yeah, I get that. I assumed, I was just sort of dumbfounded that my first reading of the post was that it was a bit and then I realized, wait a sec, this was sincere, and it had nothing to do with how you wrote it, just the culture of the internet these days
@Icebergy so you know how there are those weird people who reply to everything with a chatbot and don't realize everybody knows they are replying with a chatbot and find them cringe and annoying?
imagine if those people also wanted to fuck their chatbot
@kalomaze@leaguepublicacc Its kind of interesting how wildly different perceptions are. Like there was a bell curve of reactions to moltbook. Super technical people called bs, semi technical people were enamoured, non technical "luddites" also called bs but because ai is fake and a capitalistic ploy
@_ontologic I'm not even saying this in a "I have a good understanding myself" way. I'm just saying everybody seems to just have a lot of assumptions stemming from a weird mix of culture, religion, and intuition that they never seemed to have sat and been critical of.
@_ontologic The one thing that has surprised me over the past couple years is how confident the average person is about how consciousness and intelligence works.
I'll be real, the codex cli should just also close when you type "exit" cause I've probably spent at least a dime this month on generating "Sure, how would you like me to exit?" messages