Comment on Hype-fueling science fiction or plausible scenarios?
Thedogdrinkscoffee@lemmy.ca 4 days ago
Like all technology, the development times get shorter and shorter. No one can predict the future and specific dates are meaningless, but it is plausible that, stumbles aside, AI achieves superintelligence in the very near future. My gut says sooner than later with 90% probability by 2040. <-remember I said dates are meaningless. ;)
The world is not ready for it. It will be more destructive than constructive. May god have mercy on us all. I hope I am wrong.
taladar@sh.itjust.works 4 days ago
No, it absolutely isn’t. AI hasn’t even hit the slightest sign of actual intelligence so far so there is no reason to assume that it will get super-intelligent any time soon without some major revolutionary break-through (those are by definition unpredictable and can not be extrapolated from prior developments).
sugar_in_your_tea@sh.itjust.works 4 days ago
Exactly.
This is like the Pareto principle (80% of the benefit for 20% of the work), except the end state isn’t “general intelligence,” it’s a chat bot. There’s no general intelligence at the end of this road.
Thedogdrinkscoffee@lemmy.ca 4 days ago
I respect what you are saying, and agree our LLM are just superficial pseudo communication engines. Theories of intelligence differ, but I don’t think human intelligence, i.e. IQ of 100 is really all that far ahead of LLM. Just talk to your average person. We are easily wrong. We have bias and error. We repeat and amplify untruths and make mistakes.
It could be fewer breakthoughs than you think to get there. Not because the tech is so advanced, but because we are not that special.
Super-intelligence is just degrees better.
taladar@sh.itjust.works 4 days ago
I am not talking about degrees of intelligence at all. Measuring LLMs in IQ makes no sense because they literally have no model of the world, all they do is reproduce language by statistically analyzing how those same words appeared in the input data. That is the reason for all those inconsistencies in their output, they literally have no understanding at all of what they are saying.
An LLM e.g. can’t tell that it is inconsistent to talk about someone losing their right arm in one paragraph and then talking about them performing an activity that requires both hands in the next.
Thedogdrinkscoffee@lemmy.ca 4 days ago
I had a lot of fun talking to one about how we could fully electrify a hobby farm without batteries, by using a high voltage DC overhead wire on a cable pulley system. Everything from electrical engineering, to cable selection, sourcing vendors and costs.
Mistakes were made, but it was genuinely helpful at times. I was impressed at how far it has come.