Why are language-models so bad at math?

Question

I've interacted with a number of "sophisticated" language-models like Chat-GPT, GPT-3, Jasper and others. They all fail at the most simple math questions. Sometimes they are not even able to count a list accurately and contradict themselves when asked the same question repeatedly.I've looked at some resources to answer this question but nothing really explains why they sometimes do get the answer right or somewhat right and sometimes incredibly wrong.I'm curious to hear from people with more domain knowledge.

soueuls · Accepted Answer

Because it&rsquo;s primarily being fed a corpus of texts.There is a finite number of articles on the internet.There is an even more finite number of articles talking about Joan of Arc or copywriting.But numbers are infinite.Not many people write articles about why 2+17 is 19.Not many people write about why 33 + 4 is 37 either.