They can, they can crowd out good comments with an absolutely crushing volume of crap. While humans can put out a lot of crap, bots can do orders of magnitude more. The issue is that it is an effort multiplier for when a small number of people want to target a particular forum.
I imagine the problem isn’t so much volume of crap comments as much as the tailoring of crap comments. Imagine if every tweet-into-the-void from a human with 50 followers reliably got engaging replies. Bots taunting your grammar mistakes, bots selectively quoting your prior tweets to point out contradictions, bots cleverly insinuating that your tweets reveal problematic sympathies.
So much of our noise-filtering is ignoring comments that are too generic to be human. What happens when every spam comment seems to understand the OP, even when the OP’s true audience is negligible?
If a person has any paranoid tendencies, this would be a psychological onslaught. Interrogators use the tactics you just described to siege a person to psychological exhaustion.
Product Devs for the CCP will have a lot to work with if this ever evolves.
Would you be fooled by that? No because once it became well known, people would adjust their heuristics for judging who's real. If that's too hard, platforms would help, such as by verifying identity more thoroughly. I've never heard any realistic description of how AI bots could somehow undermine society even close to the amount that humans already do.
> What happens when every spam comment seems to understand the OP, even when the OP’s true audience is negligible?
My hope is the next step will be filtering by insightfulness/usability of a comment and then those best bots bought and used by next stack overflow: https://xkcd.com/810/
How will they get those extra comments to be seen? They still have to log in and get past captchas and all the same issues that bots have always faced. Humans create a crushing volume of crap already and we already have ways to hide that from ourselves no matter the volume.
Imagine buying a bunch of twitter accounts that have already been active for 1-2 years and then making a bunch of Tweets to influence public opinion.
I've done some experiments with GPT-2 and it had so so performance refining with tweets. Using GPT-3 you could probably just do it using only generation.
