[throwaway1004]

No doubt that is true but it doesn't seem to account for the discrepancy, at least not without acquiring more data. Google's LLM is spitting out ~250K immigrant and first-generation Irish fighting in the Civil War, sitting roughly at 10% of the total. That doesn't account for second-generation and older, so I would be interested to see what social mobility was like for those more-established Irish-Americans: were they languishing in the working class, or did they exit from the middle class to go fight in wars disproportionately?

edit: how significant is the class factor, anyway? Does that model really apply, and how many citizens would be considered middle or upper class; enough to make a large statistical impact?

[daveguy]

Google's LLM is an LLM. Results from an LLM are garbage unless you can test them. Any other sources?

[throwaway1004]

Yes, I made sure to highlight that this was the (unreliable) source, consider it an invitation for the history buffs to chime in.

Food for thought: if I had posted the very first source available on Google as authorative, with no actual knowledge of my own to make the claim, that could be more misleading on aggregate, right?

[daveguy]

Agreed, it's not so much where you find the source, but the source itself. Unquestioningly taking "first result" is just as bad as taking anything from an LLM and representing it as a factual answer. Also, kagi generally has higher quality search results (that can be checked).