Comment by mrcwinn

I feel like this is a sandboxing and ownership problem, in a sense. If everyone had personal access to AI, then we could say people are personally responsible. And we’d want to arm those people with safeguards to prevent accidental bad behavior.

I think Dario worries because he knows this isn’t about people with AI. It’s just AI for itself, running without the human, and deciding to do bad things.

Why would Anthropic build that future? Oh, I know. Enterprise revenue.

Despite the ads business and how absolutely loathsome Greg Brockman seems, at least OpenAI is seemingly focused on mostly on human beings having access to their product.

The scariest thing about Anthropic is the very thing they’re known for in a positive light: their morality. But these are the gray rules all of us navigate every day.

It’s wrong to kill, but what if killing saved ten others? Claude may has a constitution, but every evil doer acts for a “greater good.”