Auto-generated media can be a way for people to create works we can envision, but don't (yet) have the skills to realise. But in doing so, they can't help but erode our motivation to develop our artistic skills, or to reach out to those who already have, and collaborate on new creative works.
What interests me here is not the tech, the business models, or the political economy of these tools. But rather their social effects. Do they bring us together, or make it easier be alone?
Jan :rust: :ferris:
and 2 others
boosted
"The short version is that YouTube used AI to 'enhance' the videos of Rick Beato and Rhett Shull. YouTube used an AI smoothing filter that made the videos look like they were AI generated, and didn’t tell Beato, Shull, or their viewers that they were doing it. They only admitted to it after the two made videos proving it happened, later claiming it was only a 'limited test'.”
@joshsjunkdrawer, 2025
https://joshgriffiths.site/youtube-is-awful-im-not-posting-there-anymore/
(1/2)
@icedquinn
> I don't think either of the cats here have had opinions on data processing
The Trained #MOLE didn't have any opinions on it either. It just regurgitated some words vaguely related to the words you prompted it with. Any opinions involved were your own.
> where sonnet at least confirmed there was nothing else for me to learn about the problem
Rubber duck debugging at work ; )
"... a wide variety of “theories of everything” that have arisen from non-expert humans prompting large language models (LLMs) with a repeated series of inquiries, where the LLM will agree that the human’s ideas have merit and will fawningly tell the human that they’re truly onto something profound.
This is not 'vibe science' the way one might be able to 'vibe code' ..."
#EthanSiegel, 2025
https://bigthink.com/starts-with-a-bang/10-scientific-truths-unpopular-2025/
No it absolutely is! For exactly the same reasons.
@adwright
> One may only imagine my surprise at the news Bluesky is pulling posts critical of ICE and the regime
Link please? I couldn't find anything relevant in a quick web search (probably thanks to the fact that all web search is dogshit now thanks to #MOLE Training slop).
Assuming this is accurate (I have no reason to assume it's not), who does it affect? Everyone in the ATmosphere? Everyone whose PDS is hosted by BS Inc? Everyone who hasn't opted out of BS moderation?
"The company launched Workspace Studio on December 3, 2025, putting AI agent creation directly inside Gmail, Docs, and Drive."
#MarcusSchuler, 2025
https://www.implicator.ai/meta-poaches-apple-exhales-google-divides/
Well that tears it. When I'm back in the studio in the new calendar year, I'm going to prioritise finishing the deGoogling I started years ago. I have one GMail address I'm still actively using. Not for much longer.
@drwho
> many people don't work to identify and yoke them
You're talking about their personal Dark Side, and I agree with your sentiments.
But funnily enough, the whole train of thought was triggered by a comment Arden Leigh made about chatbots when she was interviewed for Team Human. She raised the possibility that of there are noncorporeal entities looking for a way in to our reality, a Trained #MOLE could make an excellent vessel.
Douglas @Rushkoff speculates about a process where the wealthy sandbox themselves, employing Trained #MOLE workers, and pay them in crypto datacentre vouchers that can be redeemed for compute, memory, storage, and so on. The end of capitalism as the Rapture of the Nerds;
https://www.teamhuman.fm/episodes/342-the-joy-of-being-worthless-except-to-each-other
Intriguing, but it reminds me of the IR realist friend who asked me "how many armoured divisions does BitCoin have?".
Anyway, I'm looking forward to seeing all the hardened prohibitionists who attack drug law reform efforts - only because they're concerned about mental health, of course - these knee-jerk conservatives, publicans, newspaper editors and alcohol industry lobbyists, all lobbying to ban generative models;
Shock! Horror! AI causes psychosis!?!
They're not going to ignore the same risk when it comes from a different source, are they? : P
(3/3)
"Another case involved a man with a history of a psychotic disorder falling in love with an AI chatbot and then seeking revenge because he believed the AI entity was killed by OpenAI. This led to an encounter with the police in which he was shot and killed."
#MarlynnWei M.D., J.D, 2025
https://www.psychologytoday.com/us/blog/urban-survival/202507/the-emerging-problem-of-ai-psychosis
Holy spitballs!
Prohibitionists;
Shock! Horror! ChatGPT uses can cause DEATH!?!
I note that this interview is from Oct, so maybe @pluralistic has modified his view on this since then but ...
One of the few things I really dislike among Cory's regular talking points is the misrepresentation of Shoshana Zuboff and the makers of The Social Dilemma, as if they ascribe to DataFarmers the invention of a "mind control ray";
https://www.teamhuman.fm/episodes/339-cory-doctorow-enshittification-is-not-inevitable
Some credulous, Kara Swisher style faux-critics may have made this claim, Zuboff and TSD didn't.
(1/2)
If you don't believe that ongoing interaction with proprietary digital systems has any power to shift people's thinking in unhealthy ways, you're not paying attention;
https://www.psychologytoday.com/us/blog/urban-survival/202507/the-emerging-problem-of-ai-psychosis
Which would be very unusual for Cory, so I'm puzzled as to why he keeps pushing this talking point.
Maybe it's his case for why people need to read Chokepoint Capitalism or Enshittification if they've already read Zuboff's book? They do, but there are better ways to make that argument.
(2/2)
James Cameron Banned Generative AI Use in ‘Avatar: Fire and Ash’: ‘We Honor and Celebrate Actors. We Don’t Replace Actors’
https://variety.com/2025/film/news/james-cameron-bans-ai-avatar-movies-fire-and-ash-1236597750/?utm_source=flipboard&utm_medium=activitypub
Posted into Film @film-Variety
(1/?)
@xurizaemon
> I love these archives, and hate the extraction machines that will frack them to dust for data
The #MOLE Training bubble will burst. Probably sooner rather than later;
https://substack.com/inbox/post/179453867
The digital commons will continue. Including Flickr, which was saved from deletion by the corporate zombies at Verizon who got it with Yahoo, and there's now a not-for-profit Flickr Foundation protecting its collection into the future;
https://revolution.social/episodes/our-mission-is-to-keep-flickr-pictures-visible-for/
@tuban_muzuru
> Okay, a truism would be an axiom of some sort
Ah, gotcha. Almost like a tautology; generative models spit up media that resembles blended pasts, because they work by blending up past media.
I was thinking of truism more in the sense of a cliché, something we all know. It's my observation that outside geeks circles, most people are still unaware of how #MOLE Training works, so for them Curtis' statement (especially in the context of the interview) is a novel insight.
It just occurred to me that it might be useful to create a directory of websites that are verified as created entirely by authentic humans. No billionaire-funded reputation laundering, and no use of a Trained #MOLE to vomit up "content".
Sites listed on the directory could display the logo of the directory on their website, with a link to their entry. Kind of like an organic or fair trade certification. The logo of the directory could be registered as a trademark, and protected from misuse.
@strypey I just did a test if my ears can detect ai slop. Searched on yt and there is a lot of them.
Result: they are actually pretty hard to tell 😬
And why I like foreign music , I actually don't know myself. Maybe most music here is like love , love and love.
@Stomata
> I just did a test if my ears can detect ai slop. Searched on yt and there is a lot of them
Yup. Some of it is even conceptually clever (Nirvana's Nevermind reimagined as 60s psychedelia), but IMHO it lacks a human spark.
> they are actually pretty hard to tell
I haven't done any blind listening tests yet, but I think I can tell. I've listened to a *lot* of music (recorded and live) over a lot of years, and played in a few banda. There's a kind of flatness to Trained #MOLE music.
@strypey I dont have the headspace to get into your meta analysis here
... however, I fancy lobbing at you some weird clip from Russian television programming I came across (which is equally above my paygrade):
https://www.reddit.com/r/lazerpig/comments/1p51r8i/nothing_tells_more_about_moscovia_than_russian
@indieterminacy
> some weird clip from Russian television programming
I have zero context for this. Could be real. Could be real TV footage with made up translation text. Could be generated by a Trained #MOLE. Who knows? The fact it's posted on dReddit with no reference to its source inclines me to assume the latter.