AGI will almost certainly exist within a decade. It should probably be humanity's top priority to do as much as possible to ensure the best outcome by: - minimizing risks such as weaponization and job loss financial risks - accelerating progress if we expect a good outcome

@istoica05 open source makes much more strategic sense for china because they are not in the lead. being in the lead in open source is not being in the lead in ai, the us industry has significantly better models. worrying about losing the lead is an argument against open source, not for it.

@slow_developer xAI doesn't have a model that surpasses deepseek. Anthropic might have a model better than o3.

@deanwball it's much more than an inconvenience for many. even if it's more likely to be a net benefit for humanity, it's still legitimately scary and anyone who isn't worried at all probably doesn't fully grasp the situation.

finding a good alternative could be hard cause agi is just inherently a very dangerous thing. it means we as a society need to get on this and do whatever we can to come up with better alternatives and maybe also work on ways to make open source safer and not ignore the risks.

@tegmark "All my doomer friends are doomers"

@_xjdr Jepa? Was it that architecture inspired by the feline brain

Ok I get it you like deepseek now can you all shut up? Also open source bad (timelines are short anyway and oss is extremely dangerous, we should strive for a less crappy option)

@ggerganov Exactly as Gary marcus predicted

@MarcThatcher @mattecapu What? Is that for real cause it sounds like fake news

@rhizosage @7oponaut I don't think trump perceives tiktok as a national security issue, but he probably does think that ai is a national security issue. Idk if he actually cares about national security though, I'm not in his head.

@ClementDelangue Very irresponsible especially considering timelines are already short anyway. Open sourcing the most powerful weapon we can imagine is extremely dangerous (and no I don't think deepseek is dangerous but it leads to dangerous future tech)

@giffmana I think the first dip was when deepseek dropped mla and the second one was after a tweet from Gary marcus saying ai is hitting a wall iirc

Concerning, but can't say it's surprising

Just because elon is a nazi doesn't mean he did the nazi salute on purpose. After all he is also an idiot so we should give him the benefit of the doubt.

Honestly I don't know why anyone would just assume good faith from grifty at this point. I don't know how likely it is that he would move the goalposts as far as they can be moved but still somehow wouldn't realize he's doing that and also argue with people pointing it out.

After accusing lecun in the past of being a grifter, now grifty griftus the clown accuses the researchers who worked on what is currently the most advanced ai in the world of being grifters. Unbelievable that anyone takes this clown seriously.

@christinexye did you try vibe check?

@xwang_lk many would be willing to pay for better search

I like bengio, my impression of him is that he's way smarter than the average doomer. it's sad to seem him spouting the usual doomerism motivated reasoning nonsense.