I was sorta aware of things like this already when I wrote the OP in 2021, but only sorta; mostly I was reasoning from first principles about what LLM technology would enable. I think I was mostly wrong; censorship hasn’t progressed as quickly as I forecast in this story. I think… I don’t actually know what the social media companies setups are. But e.g. according to the above article:
Then there is the work going on at the National Science Foundation, a government agency that funds research in universities and private institutions. The NSF has its own program called the Convergence Accelerator Track F, which is helping to incubate a dozen automated disinformation-detection technologies explicitly designed to monitor issues like “vaccine hesitancy and electoral skepticism.”
...
In March, the NSF’s chief information officer, Dorothy Aronson, announced that the agency was “building a set of use cases” to explore how it could employ ChatGPT, the AI language model capable of a reasonable simulation of human speech, to further automate the production and dissemination of state propaganda.
So if they are just talking about integrating ChatGPT into it now, they probably haven’t integrated lesser LLMs either, and generally speaking the integration of new AI tech into censorship, recommendation algorithms, etc. is proceeding more slowly than I forecast.
This longform article contains a ton of tidbits of info justifying the conclusion that censorship in the USA has been indeed increasing since at least 2016 or so, and is generally more severe and intentional/coordinated than most people seem to believe. https://www.tabletmag.com/sections/news/articles/guide-understanding-hoax-century-thirteen-ways-looking-disinformation#democracy
I was sorta aware of things like this already when I wrote the OP in 2021, but only sorta; mostly I was reasoning from first principles about what LLM technology would enable. I think I was mostly wrong; censorship hasn’t progressed as quickly as I forecast in this story. I think… I don’t actually know what the social media companies setups are. But e.g. according to the above article:
So if they are just talking about integrating ChatGPT into it now, they probably haven’t integrated lesser LLMs either, and generally speaking the integration of new AI tech into censorship, recommendation algorithms, etc. is proceeding more slowly than I forecast.