Someone is wrong on the internet (AGI Doom edition)
What can LLMs never do?
strangeloopcanon.com
LeCun points to four essential characteristics of human intelligence that current AI systems, including LLMs, can’t replicate: reasoning, planning, persistent memory, and understanding the physical world. He stresses that LLMs’ reliance on textual data severely limits their understanding of reality: “We’re easily fooled into thinking they are intel... See more
Azeem Azhar • 🧠 AI’s $100bn question: The scaling ceiling
In particular, there is no such thing as “general” intelligence. On an abstract level, we know this for a fact via the “no free lunch” theorem — stating that no problem-solving algorithm can outperform random chance across all possible problems. If intelligence is a problem-solving algorithm, then it can only be understood with respect to a specifi... See more
Link
But what physicists do so well, and most of us do so poorly, is that they carefully delimit what Newtonian and Einsteinian physics are able to explain. They know down to many decimal places where those maps are useful guides to reality, and where they aren’t. And when they hit uncharted territory, like quantum mechanics, they explore it carefully i
... See more