> New paper: LLMs are useless, biased and will make your little kid cry
> Looks inside
> LLMs used in the study: 4o, sonnet 3.7
Every. Fucking. Time.
GLM-5.2 might hurt the bottom line of Anthropic and OpenAI. Probably that's why they released its weights.
One of the things the neo-Luddite Left really needs to hear a lot more is the Global South *loves* AI.
So many people there see it as a an equalizer that makes them more able to compete in the global economy against those of us from wealthy countries.
I just learnt about approval voting and decided that it would solve all of the problems of modern societies.
Let's ignore the AI race and Meta's failed attempt to participate for a moment. What is the future of Facebook?
Vibes are seriously off. Fb is hostile both to content creators and users. Most people are there because of inertia/network effect, but it's not clear how much Meta can get away with.
Facebook is not at that point yet. It can probably still be saved. Maybe it's not even struggling and I'm just having an incorrect mental model about their operations and their future.
But from down there the optics is not good.
Ig seems to be in a better place, but it was also affected by Meta's recent authentication bug.
With these huge SM sites that are held together by the network effect, there is no point when the decline is obvious but can still be reversed. When it becomes obvious for everyone, it's already over.
Ig seems to be in a better place, but it was also affected by Meta's recent authentication bug.
With these huge SM sites that are held together by the network effect, there is no point when the decline is obvious but can still be reversed. When it becomes obvious for everyone, it's already over.
I'm curious how long China will allow releasing open weight models.
They probably have an incentive to provide open weight models close to the level to public closed American SoTA because that lowers the margins of American labs.
