Deep learning models can create content as text (e.g. GPT) and images (e.g. DALL-E), and may soon also generate audiovisual content.
Will there be any large-scale protests (>1000 people gathered in any physical location) about content generated by Deep Learning by 31st Dec 2025?
This could, for example, take the form of strikes about labour automation, protests against particular types of generated content, or protests about a news story which is fabricated by a deep learning model.
1,000
3.00@JordanSchneider Good question, and hard to be clear/firm about fuzzy things like this. I think, roughly speaking, I'd probably make a judgement call about how much of the protest was about AI content (e.g. one of 3 main talking points -> 30%) and multiply that factor by the size of the protest.
Betting based on my feeling here, but deep-learning generated audiovisual content doesn't strike me as the kind of thing people organize large-scale protests around. It doesn't physically hurt people or impede on bodily autonomy, and it doesn't directly involve the government or politics. While it might indeed have an impact on our information diet, it's hard to see that happening without the people who are consuming the content actually liking the new content. So it seems the main people who would be angry about this would be the unemployed content creators, but it's not clear to me that there would be enough in one place to meet the >1000 threshold.
@BoltonBailey Have you been following ChatGPT? I could see the culture war evolving into the war over LLM output and culture war is a great way to get protests.
@WarlockTiny I have been following ChatGPT, but it's hard for me to see how a culture war over LLMs could play out. I feel like some things in the ballpark of "tech stories in the news that people get mad about" could be
AI in the Justice system
Crypto scams
Big tech companies harvesting data
Yet I'm not aware of any 1000 person protest in the last decade that I feel has been about any of these things per se.
@ZiquaftyNny Presumably this would not count, as murderbots do not "generate content". @SamBrown, would you confirm this?
@BoltonBailey I wouldn't consider e.g. autonomous lethal weapons to count, by themselves. But if an LLM was outputting instructions on how to build weapons, that would.