Daniela Tafani <daniela.taf...@unipi.it> writes: > Melanie Mitchell, Can Large Language Models reason? > > https://aiguide.substack.com/p/can-large-language-models-reason
disponibili anche qui: https://web.archive.org/web/20230911043145/https://aiguide.substack.com/p/can-large-language-models-reason e qui: https://archive.ph/TVsNy > (non lo copio di seguito perchè immagini e formattazione servono) si ma potrebbe essere utile un po' di testo sia a fini archivistici che per aiutare gli iscritti a capire se vale la pena approfondire o meno --8<---------------cut here---------------start------------->8--- [...] Reasoning is a central aspect of human intelligence, and robust domain-independent reasoning abilities have long been a key goal for AI systems. While large language models (LLMs) are not explicitly trained to reason, they have exhibited “emergent” behaviors that sometimes look like reasoning. But are these behaviors actually driven by true abstract reasoning abilities, or by some other less robust and generalizable mechanism—for example, by memorizing their training data and later matching patterns in a given problem to those found in training data? Why does this matter? If robust general-purpose reasoning abilities have emerged in LLMs, this bolsters the claim that such systems are an important step on the way to trustworthy general intelligence. On the other hand, if LLMs rely primarily on memorization and pattern-matching rather than true reasoning, then they will not be generalizable—we can’t trust them to perform well on “out of distribution” tasks, those that are not sufficiently similar to tasks they’ve seen in the training data. * What Is “Reasoning”? The word “reasoning” is an umbrella term that includes abilities for deduction, induction, abduction, analogy, common sense, and other “rational” or systematic methods for solving problems. Reasoning is often a process that involves composing multiple steps of inference. Reasoning is typically thought to require abstraction—that is, the capacity to reason is not limited to a particular example, but is more general. If I can reason about addition, I can not only solve 23+37, but any addition problem that comes my way. If I learn to add in base 10 and also learn about other number bases, my reasoning abilities allow me to quickly learn to add in any other base. * “Chain of Thought” Reasoning in LLMs In the last few years, there has been a deluge of papers making claims for reasoning abilities in LLMs (Huang & Chang give one recent survey). One of the most influential such papers (by Wei et al. from Google Research) proposed that so-called “Chain of Thought” (CoT) prompting elicits sophisticated reasoning abilities in these models. A CoT prompt gives one or more examples of a problem and the reasoning steps needed to solve it, and then poses a new problem. [...] * Are the Reasoning Steps Generated Under CoT Prompting Faithful to the Actual Reasoning Process? [...] * If LLMs Are Not Reasoning, What Are They Doing? [...] * The Trickiness of Evaluating LLMs for General Abilities [...] [...] One might argue that humans also rely on memorization and pattern-matching when performing reasoning tasks. Many psychological studies have shown that people are better at reasoning about familiar than unfamiliar situations; one group of AI researchers argued that the same patterns of “content effects” affect both humans and LLMs. However, it is also known that humans are (at least in some cases) capable of abstract, content-independent reasoning, if given the time and incentive to do so, and moreover we are able to adapt our understanding of what we have learned to wholly new situations. Whether LLMs have such general abstract-reasoning capacities, elicited through prompting tricks, scratchpads, or other external enhancements, still needs to be systematically demonstrated. --8<---------------cut here---------------end--------------->8--- Cordiali saluti, 380° -- 380° (Giovanni Biscuolo public alter ego) «Noi, incompetenti come siamo, non abbiamo alcun titolo per suggerire alcunché» Disinformation flourishes because many people care deeply about injustice but very few check the facts. Ask me about <https://stallmansupport.org>.
signature.asc
Description: PGP signature
_______________________________________________ nexa mailing list nexa@server-nexa.polito.it https://server-nexa.polito.it/cgi-bin/mailman/listinfo/nexa
