Chatbots can sway political opinions but are 'substantially' inaccurate: study
theguardian.com
theguardian.com
It makes sense that LLMs would want to ape that.
But does misinformation form people’s opinions or exacerbate them. We’re pizzagate / Qanon folks changed by that misinformation or were they looking for an excuse?
I think the amount that exacerbates political opinions is degrees.
I certainly laugh at certain false memes making fun of the other side, and there are probably facts I don’t check because they inform my priors, but I don’t know that misinformation changes my political opinions all that much, in fact if I feel that my first instinct is to do a fact check.
They're arguably a special case, in that qanon was primarily a fantastical conspiracy theory along the lines of the blood libel or the satanic ritual abuse thing, and arguably only incidentally political. People who end up getting into that sort of thing are an edgecase. They are not simply believing disinformation; they are believing in _magic_.
But RE more _conventional_ misinformation, IIRC there was polling showing, for instance, that a substantial minority of British people believed various common 'euromyths'. Since the 70s or so, British tabloid media has pushed various bizarre claims about European legislation, and a lot of people seemed quite happy to believe, for instance, that Europe had banned jam, even though jam was still clearly available in the supermarket. Many people even believed that Europe had banned or had tried to ban sausages, apparently due to misremembering a joke in Yes Minister as actual news. Given how tight the Brexit vote was, it's fairly plausible that euromyths swung it.
I do observe that there are some folks who seem more vulnerable to misinformation than myself, and I do think it's a debate worth having.