3 ms·

I hold similar opinions. I find even the flaws that the latest attention based language models have to be interesting in the way they fail... by lying, making s

by kurthr 4y ago

I hold similar opinions. I find even the flaws that the latest attention based language models have to be interesting in the way they fail... by lying, making stuff up, and stating BS with certainty. Ask why they answer the way they do and it's more confident truthy sounding gibberish. That doesn't make these conscious by any measure, but it rhymes. Were only so much rational animals, but fully rationalizing ones.

ako 4y ago
They’re not aware when they’re failing, they just repeat the data they’ve found. These models are aware they’re lying or BSing. Rubbish in, rubbish out.
- tsukikage 4y ago
  > These models are aware they’re lying That's a fascinating statement. Can you expand on that? Are you suggesting there is something in there that can be aware of things? What does being aware even mean?
  - ako 4y ago
    My mistake, typed too fast, should have had a not in that statement.