If an LLM hallucinates in 1% of occasions and gives subpar output in 5%, this kills his effectiveness to replace anyone. Imagine a support guy on the other side of the phone to speak gibberish 10 times a day. Now, imagine a doctor. These will never lose their jobs.
but llm's dont speak gibberish 10 times a day even now. from my usage, chatgpt has not said one obviously strange thing since o3 came out.
> Imagine a support guy on the other side of the phone to speak gibberish 10 times a day.
A massive improvement?