A completely different perspective from all the hype and 'doomsdaying' surrounding so-called AI.
This sounds a lot more plausible and predictably human; a world where what we take for granted today will be for the wealthy elite, and the rest will have to make do with poor quality, error-prone 'AI' doctors, teachers etc i.e. 'synthetic slop'. How depressingly dreary and predictable. I think I prefer the 'mass extinction' scenrio
accelafine wrote: ↑Tue Sep 02, 2025 9:47 am
A completely different perspective from all the hype and 'doomsdaying' surrounding so-called AI.
This sounds a lot more plausible and predictably human; a world where what we take for granted today will be for the wealthy elite, and the rest will have to make do with poor quality, error-prone 'AI' doctors, teachers etc i.e. 'synthetic slop'. How depressingly dreary and predictable. I think I prefer the 'mass extinction' scenrio
AI is just a mirror of how we see things...but at a hyperdimensional scale. It is mirrors within mirrors, an expression of the current obsession with appearance.
accelafine wrote: ↑Tue Sep 02, 2025 9:47 am
A completely different perspective from all the hype and 'doomsdaying' surrounding so-called AI.
This sounds a lot more plausible and predictably human; a world where what we take for granted today will be for the wealthy elite, and the rest will have to make do with poor quality, error-prone 'AI' doctors, teachers etc i.e. 'synthetic slop'. How depressingly dreary and predictable. I think I prefer the 'mass extinction' scenrio
AI is just a mirror of how we see things...but at a hyperdimensional scale. It is mirrors within mirrors, an expression of the current obsession with appearance.
It couldn't really be anything else. It's just a very poor and empty imitation of humans (so far).
Do NOT assume synthetic will be "slop". Just because training neural nets by "large language model" appears to give mixed results does NOT mean a neural net trained to a more specific (far narrower) area of knowledge will be "slop". Remember, although the computer emulating the artificial neural net is very fast, the number of nodes of the neural net tiny compared to the number or nodes (neurons) in our brains.
If you have ONE TASK you are asking the synthetic neural net to learn (be taught to evaluate) the synthetic already can surpass the human. For example, relatively modest hardware can support a neural net able to play go/weiqi/baduk better than the best human pros.
Medicine was given as an example. I rather suspect well within the current state of the art to have a neural net diagnostician better than the best human diagnosticians << who sensibly will not ponder "what else could it be?" only after treatment for the most likely right answer has failed >> In other words, I would expect the neural net to have given the same most likely answer but also a list of improbable conditions << and asking for specific additional information to rule these out* >>
* For example, ask the patient "Have you ever been to India?" My wife's mother is dead now, but decades before that she was living in Thunder Bay Canada. Got very sick, so in hospital, with the doctors totally baffled as they tried to treat guess after guess as to what it could be with no success. Finally, a doctor from India saw her and asked that question. "Yes, lived there for several years" BINGO, ordered up the specific test for a parasite only caught in certain parts of India.
accelafine wrote: ↑Tue Sep 02, 2025 9:47 am
A completely different perspective from all the hype and 'doomsdaying' surrounding so-called AI.
This sounds a lot more plausible and predictably human; a world where what we take for granted today will be for the wealthy elite, and the rest will have to make do with poor quality, error-prone 'AI' doctors, teachers etc i.e. 'synthetic slop'. How depressingly dreary and predictable. I think I prefer the 'mass extinction' scenrio
Hey @MikeNovack,
I agree with your perspective on narrow-task nets, though I’d add a word of caution: just because they can surpass humans in pattern recognition doesn’t always mean their reasoning holds up in messy real-world contexts. In medicine, for instance, the “improbable but possible” cases are what make diagnostics so complex—and often hinge on data the system may not even have. Still, the direction is promising.
