5/2/2026 at 3:53:30 PM
“The central problem is that LLMs are, by design, nothing more than statistical pattern matchers, making probabilistic guesses to answers”Any time this quote pops up, credibility goes to zero. It’s virtually the most surface level understanding of what AI is and also what human intelligence is. Quoting outdated METR study (n=20, model=sonnet 3.x), claiming that no meaningful progress has happened in last couple of years (and then their citation footnote which I was pleased to see actually does not have anything to do with their statement).
Claiming Uber was basically a scam, running at a loss and investors just trying to cash out as soon as possible in an IPO (go look up the market cap of Uber today and tell me their business model was a scam).
But people eat this stuff up?
by aspenmartin