gfodor.id@gfodorNobody would be talking about R1 if DeepSeek didn’t dump the weights imo. This is why it’s especially cringe: the “Sputnik moment” happened because OpenAI abandoned their core mission. That’s the root cause.Opens with an observation
gfodor.id@gfodoro3 has this insane attribute where it has the knowledge and intellect of a PhD level grad student but also is scrappy and wants to get stuff done. These two combinations together in an individual, at the same time, is excruciatingly rare. And now it’s on tap in a model. wtfOpens with an observation
gfodor.id@gfodorI hate to break it to everyone, but America First is going to literally be the policy for domestic AI models going forward. Now that Dario woke up the sleeping giant and they realize the lever they have here for US economic dominance you can be certain they will keep smashing it.Opens with an observation
gfodor.id@gfodorI’m torn, because I think frontier labs have a moral duty to open source models once it’s safe to do so. DeepSeek stealing from Anthropic, a company that hasn’t open sourced a damn thing, to release a frontier model to the world are closer to Robin Hood than common thieves.Opens with an observation
gfodor.id@gfodorOk so apparently the plan now is to use the fancy term “distill” to lock people into this dumb narrative that including some raw OpenAI API slop in your training set (if they did that) is some kind of profound IP theft. So embarrassing for the US. Make it stop.Opens with an observation