On: Sam Altman: 'I No Longer Think We're Gonna Be Able to Just Patch All the Bugs'

[eddythompson80]

Yes, I'm. I didn't mean for my list of extremes to be an exhaustive list. But yes, NOBODY (in a "butlerian jihad" sense) is just another example in the list of extreme takes. Only governments or military is another extreme.

[za_creature → eddythompson80]

There are only two extremes: no regulation or a complete ban.

You only mentioned one of them, and claimed that the current status quo is the other. I find that disingenuous.

edit: current status quo is "best models are only for internal use", broadly equivalent to "nobody except openai and antrophic"

[eddythompson80 → za_creature]

Nope, you are not really comprehending my comment, and that’s a false dichotomy. Here is a non-exhaustive list of extremes:

- No regulations what so ever, it's free for all

- Complete ban on all AI

- Only OpenAI and Anthropic can do AI

- Only the Government, the Military, NASA or other entities approved explicitly by Congress can do AI.

That's not the current status quo either

[za_creature → eddythompson80]

You cannot have less regulation than "no regulation" and you cannot have more regulation than "completely forbidden", which are the true extrema.

A complete ban for all is obviously _more_ restrictive than a ban on entities explicitly approved by the legislative, it doesn't really matter if those entities are state owned or private sector since none < some

To make myself perfectly clear, I am generally in favor of a complete ban on any non-cryptographic algorithm with provably non-deterministic outcome (same data goes in, meaningfully different data comes out each time).

[eddythompson80 → za_creature]

Completely incoherent reasoning that's falling for a clear false dichotomy fallacy.