I only scrolled over it but every time I stopped to read another sentence, it got worse.
From what I understood open-weights models are too efficient, too cheap, don’t consume enough energy, and don’t require to construct as many super data centers. That’s why they are… dangerous?
The schizophrenia of the AI industry is laid bare in this fever dream of a statement. My god, we really cannot afford to have tech CEOs running things as a society. All they seem to be interested in is killing us.
I think the two principle arguements he’s making are open weights allow for misaligned models and these models are using distillation to more efficiently train.
I don’t really care about the distillation point - it seems it’s just hoovering up the internet with extra steps. It’s really up to Anthropic to deal with by enforcing their ToS. Sauce for the goose etc…
I can see the argument about alignment but it’s not a new problem. In the FLOSS world we generally don’t put restrictions on use on the code we release even if it might get used for bad stuff. It certainly wouldn’t hurt for the authorities to test open models to find out where the guardrails are but ultimately rather than banning the tech you need to enforce the law against the people that use the tech to do bad stuff.
Fuck “alignment.” If I want to make an AI dungeon master, it’s almost impossible to do so (at least the way I enjoy tabletop RPGs) because the model can’t promote violence or other sorts of evil.
I get it, if you’re making a chatbot, you don’t want it suggesting wanton murder to fifth graders and grandmothers. But not every valid use of AI falls into the cleanly filtered worldview of Anthropic or OpenAI.
Or maybe I have it backwards. Maybe in the future everyone will flock to porn and ultraviolence because it’s the only way to find genuine humanity in an AI-sanitized world.
I only scrolled over it but every time I stopped to read another sentence, it got worse.
From what I understood open-weights models are too efficient, too cheap, don’t consume enough energy, and don’t require to construct as many super data centers. That’s why they are… dangerous?
The schizophrenia of the AI industry is laid bare in this fever dream of a statement. My god, we really cannot afford to have tech CEOs running things as a society. All they seem to be interested in is killing us.
I think the two principle arguements he’s making are open weights allow for misaligned models and these models are using distillation to more efficiently train.
I don’t really care about the distillation point - it seems it’s just hoovering up the internet with extra steps. It’s really up to Anthropic to deal with by enforcing their ToS. Sauce for the goose etc…
I can see the argument about alignment but it’s not a new problem. In the FLOSS world we generally don’t put restrictions on use on the code we release even if it might get used for bad stuff. It certainly wouldn’t hurt for the authorities to test open models to find out where the guardrails are but ultimately rather than banning the tech you need to enforce the law against the people that use the tech to do bad stuff.
Fuck “alignment.” If I want to make an AI dungeon master, it’s almost impossible to do so (at least the way I enjoy tabletop RPGs) because the model can’t promote violence or other sorts of evil.
I get it, if you’re making a chatbot, you don’t want it suggesting wanton murder to fifth graders and grandmothers. But not every valid use of AI falls into the cleanly filtered worldview of Anthropic or OpenAI.
Or maybe I have it backwards. Maybe in the future everyone will flock to porn and ultraviolence because it’s the only way to find genuine humanity in an AI-sanitized world.