Remarkable chart in a new paper out of MIT looking at cognitive performance of LLM users in an essay writing task - https://arxiv.org/abs/2506.08872
@hailey "Over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels." what a quote
The abstract says
> LLM users also struggled to
> accurately quote their own work
I'm thinking (!) that a big part of the problem here is that it's not actually "their own work". It is essentially plagiarism.
We can certainly pretend that using an LLM is not plagiarism, but we will be forced to admit that the two are very similar, like in this case.
When you plagiarise, you have a hard time quoting the resulting text, since you did not write it. Similar to LLM usage.
1/2
@kirakira @hailey more important than quoting your own work is *understanding* your own work.
Of course, an LLM user will not understand what the text that comes out really means, understand the thought behind it, because there is no thought behind it. There is nothing to understand, there is only "content" which does not really have any meaning.
2/2
