AINext-token prediction and the debate over AI reasoning
One commenter argues that symbolic systems around transformer-based language models can soften their limitations without changing their statistical, next-token-prediction core.
TLDR
One commenter argues that every transformer-based language model fundamentally predicts the next token. Symbolic systems surrounding a model can soften its limitations, they say, but it remains a nondeterministic statistical engine. A reply offers a pointed brain analogy: “human brains are just electrical impulse emitters, they don't really ‘reason’.”
Next-token prediction and the debate over AI reasoning
One commenter argues that symbolic systems around transformer-based language models can soften their limitations without changing their statistical, next-token-prediction core.
TLDR
One commenter argues that every transformer-based language model fundamentally predicts the next token. Symbolic systems surrounding a model can soften its limitations, they say, but it remains a nondeterministic statistical engine. A reply offers a pointed brain analogy: “human brains are just electrical impulse emitters, they don't really ‘reason’.”