RT @matthiassamwald
Large language models seem to get better each time you browse Twitter.

But how sure are we they get better at things *that matter*?

https://doi.org/10.1016/j.jbi.2022.104274

1/