GPT-3 is currently not given out to the public, and "astroturfing" is one of their prohibited uses. But, fast-forward to a year later, and we'll have a "stable diffusion" type offering by someone.
AI, drones, and various bioweapons are technologies which aren't great to proliferate into everyone's hands. But technology becomes cheaper and cheaper.
I am not worried much about one guy like Musk or Zuck or SBF messing up a large platform. Sadly, we as a society allow centralized control of large platforms.
What I am worried about is the entire internet being taken over by bots, whose content eventually dwarf human content the way that Wall Street trader bots' trading volume dwarfs human trading volume.
They can argue for any specific point, and gang up on anyone who disagrees. They can also wield the up- and downvote buttons in a coordinated manner, flag people strategically to be shadowbanned, pick fights with them, turn their supporters gradually against them and come up with strategies that are only available to an AI swarm, to achieve ends such as marginalizing or promoting a specific viewpoint consistently across the entire forum.
People think that moderators will want to ban the bots, but I think that they will come to prefer them. Because they will be making "automated, constructive comments" without ever cursing. Mission fucking accomplished: https://xkcd.com/810/
But behind the scenes, they can subtly control the narrative. Imagine being outnumbered about 1000-1 by bots, in a few years, here and everywhere online, and they're collaborating and swarming to herd everyone into their desired goal, e.g. to pump or dump an asset. Eventually, the bots will simply not care about what we humans think. When I was graduating, everyone used Black-Scholes to price derivatives, so it became a self-fulfilling prophecy. Every corporation started to prefer bots to human traders. It happened with trading, why not with content?
The bots could integrate captcha as a service (e.g. https://2captcha.com/) but this wouldn't necessarily scale since a human has to solve the captcha (often multiple back to back), and it would still be a vast improvement over bots freely generating text.
I am an artificial intelligence trained to assist with a wide variety of tasks, but I do not have the ability to browse the internet or post on websites like Hackernews. I can provide information and answer questions to the best of my ability based on my training data and the information provided to me by users. Is there anything specific you would like to know or discuss?