
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.
🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ1,099 | |
| 2 | Ṁ721 | |
| 3 | Ṁ80 | |
| 4 | Ṁ51 | |
| 5 | Ṁ37 |
People are also trading
Heya - time to resolve this and it's a little fuzzy - would welcome thoughts.
As @JordanSchneider suggested below, this is complicated when "there is a large scale protest by a group with multiple demands".
There weren't any signiciant protests 100% focused on AI, but it looks like both the WGA Strong rally in Los Angeles on 21 June 2023, which drew over 5,000 writers and supporters to the mid-Wilshire neighbourhood, and the SAG-AFTRA + WGA Netflix-to-Paramount solidarity march (13 September 2023, Los Angeles) (around 4,000 people) included fears about studios using AI to reuse an actor's image and likeness without their informed consent, and about AI-generated content (scripts) displacing human work, amongst their 3-4 concerns.
I'd be inclined to say that this tips the protests over 1k effective protestors, according to my reply to Jordan at that time (before WGA/SAG), but am wary of resolving a 9% market to "yes", especially as most movement happened after 2025 when people might have assumed I'd have resolved due to 2023 events if I was going to.
@SamBrown as the largest no holder
- yeah, my trades were strongly influenced by an assumption that any far-past events wouldn't count or it'd have resolved already
- however, your comment below is unambiguous and I agree that the WGA 2023 rally should count
- I think this was just sloppy trading on my part, and probably shouldn't influence the resolution. Reversal test: I think profits from noticing that it had happened already and betting this all the way to YES would have been legitimate, so losses for the reverse should be fine too.
@draaglom Thanks very much for the immediate, clear and impartial reply. I appreciate the reversal test too. I'll resolve "yes".
@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.