Raphaël Millière @raphaelmilliere.com · Aug 11 Can LLMs reason by analogy like humans? We investigate this question in a new paper published in the Journal of Memory and Language (link below). This was a long-running but very rewarding project. Here are a few thoughts on our methodology and main findings. 1/9
140 likes 5 replies
Replies
Can a «reasoning» LLM ever output a reasoning never seen in, either in parts or complete, in its training set? Humans certainly can.
human level not human like - that is an important distinction to keep in mind when discussion AI!
We focus on an important feature of analogical reasoning often called "re-representation" – the ability to dynamically select which features of analogs matter to make sense of the analogy (e.g. if one analog is "horse", which properties of horses does the analogy rely on?). 2/9
📌
Don't all humans reason differently? LLMs do not reason at all - they pick out the nearest answer that might or might not fit.