Comment on Generative AI's most prominent skeptic doubles down
tal@lemmy.today 10 months ago
“I’m skeptical of AI as it is currently practiced,” he said. "I think AI could have tremendous value, but LLMs are not the way there.
Ehhh. I don’t know if I’d go quite that far. I think that LLMs might be a component of a functional AGI. But just running a larger LLM model on bigger hardware is not going to suddenly barf out something that can act in the same sort of general way a human can, I agree there.
AGI is artificial GENERAL intelligence, it is not the same as generative AI.
en.wikipedia.org/…/Generative_artificial_intellig…
GAI: is a subfield of artificial intelligence that uses generative models to produce text, images, videos, or other forms of data.
en.wikipedia.org/…/Artificial_general_intelligenc…
(AGI)—sometimes called human‑level intelligence AI—is a type of artificial intelligence that would match or surpass human capabilities across virtually all cognitive tasks
The article literally mentioned AGI, which is why he commented about the relationship between LLM and AGI. I agree with him, LLM by itself will never be an AGI, but it can definitely be a part of it, especially if it takes the role similar to the human language centre.
If AGI is made of components, you could argue that it isn’t “general”. Which would be fine, as most psychologists would say the same about human intelligence
The human brain is made of components, so why not an AI brain?
Isn’t human intelligence exactly what most people mean by “general intelligence”? The real question is whether it can be created artificially.
If AGI is made of components, you could argue that it isn’t “general”.
Can you elaborate? I don’t get what you mean by this.
The idea of general intelligence (g) is that you’ve got some overall capacity that can be turned to any task. Human intelligence is probably much more like a big toolbox of skills, and though. I worry that people who get excited about AI are kidding themselves a bit as it isn’t going to be general - it’s going to be a toolbox at best: an LLM for writing, a totally different system for drawing… And at that point you’ve not got some special interesting AGI - you’ve just reinvented the idea of apps
I think you’ve misunderstood what AGI means.
ZILtoid1991@lemmy.world 10 months ago
I don’t think so, and I believe not even the current technology used for neural network simulations will bring us to AGI, yet alone LLMs.