Human-level intelligence is possible. See: humans
Do you think it's physically impossible to recreate human-level general intelligence artificially?
Do you think it's physically impossible for any intelligence to be superhuman, even a natural intelligence?
Unless you answered "yes" to either of the above, why shouldn't it be physically possible for humans to create superhuman AGI?
Is there any physical law whatsoever that even hints that intelligence should be able to create AGI at its own level, but in no way more intelligent?
I never said there was a physical theory that predicts superhuman AGI. But in contrast, there are other concepts such as time travel which are prohibited by certain theories we have. It seems very unlikely for time travel to be possible. Superhuman AGI is not prohibited from existence.
I have no idea. I have no basis for concluding it's either possible or impossible.
> Do you think it's physically impossible for any intelligence to be superhuman, even a natural intelligence?
No, I think a godlike being could materialise an entity with superhuman intelligence, or natural selection could evolve one (perhaps even on earth -- perhaps it already has!).
> Unless you answered "yes" to either of the above, why shouldn't it be physically possible for humans to create superhuman AGI?
Even assuming I believe both the above are possible that doesn't mean I think it's possible they could occur on planet earth within any meaningful time frame. I also think that interstellar travel is compatible with the laws of physics but just because we've sent humans to the moon and robots to mars we have no idea whether interstellar travel is feasible or when it might occur. For me the response to the advent of GPT is like people saying "wow, we've sent men to walk on the moon -- if we keep going people will soon be visiting mars/alpha centauri!".
There's no rational basis (scientific, logical, anything) for knowing whether AGI is more akin to mars in this analogy (and we're really quite close) or to alpha centauri (and we're no where near).
> No, I think a godlike being...
Superhuman intelligence is a reality in many fields already: chess, Go, self-driving cars at certain conditions, calculations and many more. This is not necessary to have it for everything at all.
Sorry, what? According to which scientific theories is AGI possible, and what principles are those theories based on? What testable hypotheses say that it is, and which experimental evidence has been produced in favour of AGI being possible?