4 ms·
I asked GPT for rules on 101-level French grammar. That should be well documented for someone learning from English, no? The answers were so consistently wrong
by ted_bunny 9mo ago
I asked GPT for rules on 101-level French grammar. That should be well documented for someone learning from English, no? The answers were so consistently wrong that it seemed intentional. Absolutely nothing novel asked of it. It could have quoted verbatim if it wanted to be lazy. I can't think of an easier question to give an LLM. If it's possible to "prompt wrong" a simple task that my six-year old nephew could easily do, the burden of proof is not on the people denying LLM intelligence, it's on the boosters.
- munksbeer 9mo ago> the burden of proof is not on the people denying LLM intelligence, it's on the boosters It's an impossible burden to prove. We can't even prove that any other human has sentience or is reasoning, we just evaluate the outcomes. One day the argument you're putting forward will be irrelevant, or good for theoretical discussion only. In practice I'm certain that machines will achieve human level output at some point.
- mjcl 9mo ago> machines will achieve human level output at some point Would you care to put some sort of time scale to "at some point?" Are we talking about months, years, decades, centuries?
- munksbeer 9mo agoNo real idea. It is also a very difficult thing to measure. But I think once we see it, the argument will be over. Wild guess, within 30 years.