Comment by solyin
12 hours ago
Only the labs knows the truth, but the question as to how many Americans used ChatGPT/Grok/Claude to help them decide how to vote in the most recent election is hopefully going to be something that historians will be able to unearth later.
I wonder if we’re approaching a cultural moment where people in general are doing exactly what they did when Google first became ubiquitous—search to confirm their biases—but now with LLMs instead of Google. But where Google would happily feed them some link to potential mis/disinformation, the AI is more likely to say “no, that’s not accurate, and here’s why…”
I doubt the number is low. Knowing everything I know about llms, I still fall into the same groove as I would if I was talking to a buddy online. That part is not far fetched.