3 ms·

Sounds like you believe in a conspiracy to fool people out of their money with fake AI. But have you looked at what AI could do 5 years ago vs present day? You

Sounds like you believe in a conspiracy to fool people out of their money with fake AI. But have you looked at what AI could do 5 years ago vs present day? You can't deny there is progress. It is useful in many ways without being able to replace anyone, I agree with you on that point. I rather like this outcome.

YeGoblynQueenne 3y ago
>> You can't deny there is progress. Yes, you can. Because there are no objective measures of that "progress" since there are no good metrics for language generation. It's just people going ooh and aaah and saying "look how good it is!", or just "look how big it is!".
retrocryptid 3y ago
If I believe in a conspiracy, it's one in which people are fooled out of their money by real AI. What I've seen openai derived services do: * generate text whose assertions have no basis in reality leading to requirements that AI generated text actually be checked by humans. * generate entire fake legal citations, destroying the reputations of two lawyers. * generate code I two hours of conversation I could generate myself in 20 min. (But that's not the bad bit. The idea that a non technical product manager could build something is pretty cool.) But... I have yet to see AIs take an existing code base and then modify the code base to add features or fix bugs. What I see is AIs take a code base and completely re-write it. This is a problem because rewriting large blocks of code often introduces new behaviors and bugs. So yes, there has certainly been progress. The same way some rough beast is making progress towards bethlehem.