A USA centrist is like a Canadian center-right, at least.

I would like to have your ideas on this.

I am unsure how much you are ready to commit to the "AI is ouija idea" but you have colleages at Radboud who have shown, for instance, that LLM surprisal values from (smaller) AI models correlate with reading times, EEG spikes, etc. Generative AI can be used as ouija, but it's not all ouija.

This is the postmodernist dream: any opinion can be backed by a paper if you shop long enough.

I’m surprised government data is considered more reliable, knowing how it often has major measurement issues.

Also witness the cool boxplots: