Ahh yes, the nation-state funded military-fake-book complex, destabilizing nations the world over. We really need to watch out for nation-states using incoherent technobabble!
The commenter does have a point though; there's a lot of low effort high volume spam on the internet that reinforce certain narratives that cause division, on subjects like gender identity, generational differences, favorite TV shows, wokeism, Brexit, classism, racism, etc. There's actors on both sides of those debates, not because they're particularly leaning towards one or the other, but because the division is causing political shifts and internal strife.
This has already led to the EU starting to fall apart - the UK is out of the EU, Hungary and Poland no longer meet the criteria for having backslid into totalitarianism or at the very least, no longer having a division between the legal and executive branches of law. And it's causing lines to be drawn in the US as well, with calls for secession rising. But the US has been an "us vs them" type of "democracy" for a while now, it probably didn't take much to entrench the sides more deeply.
Anyway, that's online discourse, I'm sure it extends to books too, just less obviously.
One interesting book is Active Measures: The Secret History of Disinformation and Political Warfare by Thomas Rid. It's a history of disinformation, propaganda, and related shenanigans from the Protocols of the Elders of Zion to the 2016 election, mostly focused on the Cold War.
For the most part, it's low-effort manipulations that very occasionally take hold---I can remember the anti-nuclear protests against the deployment of the Pershing II missiles in Europe---but the internet and social media have dropped the bar for "low-effort".
Yeah, not sure what the new concern here is. "Nation-states" always have had the ability to manufacture truth out of nothing, they've been doing this since before there were any computers. Why would they need the help of GPT-3? Why would they want to spam incoherent hallucinations? What would they gain out of that when they already have the ability to "manufacture consent" at scale?
It's not even a democratisation of bullshit, non-state actors also don't need GPT-3 to spam nonsense. PR and advertising is all about that and has a history of hundreds if not thousands of years. Maybe the new thing here is that people now realise lying is easy.
