Skip to main content

This is a really nice read about whether #LLMs can actually reason: https://aiguide.substack.com/p/can-large-language-models-reason
I think expecting language models to reason like the math engines maght be a bit out of range! Nice try!

Can Large Language Models Reason?

What should we believe about the reasoning abilities of today’s large language models? As the headlines above illustrate, there’s a debate raging over whether these enormous pre-trained neural networks have achieved humanlike reasoning abilities, or …
Melanie Mitchell (AI: A Guide for Thinking Humans)
#LLMs
in reply to victor tsaran

Chi Kim
Chi Kim
• •
index with 0 vs 1 as start was interesting. However, when I learned matlab which use index 1 as start, I frequently made makes because of habit. lol
in reply to Chi Kim

victor tsaran
victor tsaran
• •
@chikim Yeah, totally get it. But I thought the example was quite illustrative.
@Chi Kim
in reply to victor tsaran

Tim Kellogg
Tim Kellogg
• •
aside from if they CAN do math, it’s an enormously expensive way to do math
in reply to Tim Kellogg

victor tsaran
victor tsaran
• •
@kellogh Luckily, there are ways to deal with this, for example, by using the Program Aided vanguage approach. But yeah, LLMS were not really built to do math. Although, the minute I said this, someone probably came up with a technique!!! :)
@Tim Kellogg
⇧