Post

We could take the same dollars, develop publicly-owned models based on consensually-gathered data, made of entirely open-source and open-weights code, and enable a thousand independent, responsible innovators to spring up instead. WTF is Bernie doing?

Goofrider@goofrider.bsky.social

Spending taxpayers money to buy up 50% of AI companies' equity at the height of their bubble phase. What could possibly go wrong?

Which algorithm is "AI"? LLMs hallucinate. They don't store knowledge or reason. They store statistical relationships between words. They are essentially a lossy index, without any original sources to reference. The indices are built from unvetted and regurgitated hallucinated data. GIGO.

sure. and people routinely use them for tasks where those shortcomings aren't necessarily disqualifying. as you know.

People hit themselves in the face with hammers to be more attractive, too. Whether those shortcoming are disqualifying is a matter of opinion. To say nothing of AI psychosis and sycophancy. There's nothing "AI" does better than a human except generating slop quicker. I prefer life with less slop.

And finally, what about: - AI surveillance and false identifications - AI weapons targeting - suicides and school shooting planned with AI - AI medical misdiagnosis and treatments - AI insurance and benefit denials - AI security bugs and supply chain vulns None of these are fixed by open models.

1 repost0 likes

No replies yet.