micro-posts from across the fediverse — for long-form, see the blog
Chiang’s strongest argument doesn’t need its headline. Forget consciousness, moral agency means bearing consequences: liability, blame, exclusion. Software can’t. So selling a chatbot as a moral agent lets its makers and users shed responsibility. theatlantic.com/philosophy/2026/06/no-artificial-intelligence-is-not-conscious/687378/
This whole story feels like the cryptography export ban that Clinton relaxed, beginning in 1996, then furthermore in 2000, without which we would not have the web as we know it.
nytimes.com/2000/01/31/business/worldbusiness/IHT-us-removes-an-encryption-barrier.html
If there was a need for another reason for the European Union to wake up : anthropic.com/news/fable-mythos-access
Relying on foreign models grants the US kill switch power over European infrastructure. That’s scary even if we are allies.
Mistral is cool but it’s really not there yet.
They laughed at Galileo. They also laughed at Bozo the Clown.
Why “they laughed at Galileo” is a fallacy, not evidence — and why, in health, the gap between persuasion and proof carries a body count.
# alt: Carl Sagan and Galileo (AI generated)
TikTok is teaching a generation what ADHD is. The problem: two clinical psychologists checked the 100 most-viewed ADHD videos and found under half the symptom claims (48.7%) were accurate. The fake-science clips weren't the fringe. They were the popular ones. A thread 👇
Researchers then ran another test. 843 young adults watched the experts' best and worst picks : they rated the worst videos higher and the best ones lower. The more ADHD content they'd consumed, the more confident they were recommending it. Accuracy was not a concern.
68.5% of the inaccurate claims just described ordinary human experience. Bumping into furniture. A sweet tooth. Replaying a song. Engagement rewards the relatable and the absolute; clinical nuance ("this may not apply to everyone") is boring. Only 4% of videos included any.
And the people posting it: 80% disclosed no credentials; in a related study, just ~1.6% of #ADHD videos came from health professionals. More concerning : about half were selling something like products, coaching or had donation links.
The fair part: people aren't stupid for being here. Formal diagnosis is slow, costly, and historically missed women, girls, and adults entirely. TikTok fills a need. But the algorithm does not care about truth, it rewards engagement.
In the UEA review's ADHD case, ~3% of professional videos carried misinformation vs ~55% from non-professionals. The fix isn't telling people to log off. It's getting clinicians into the algo.
Sources: Karasavva et al., PLOS ONE 2025 (the #ADHD study): journals.plos.org/plosone/article?id=10.1371/journal.pone.0319335 · Carter/Chatburn et al., systematic review, J. Social Media Research 2026: jsomer.org/index.php/pub/article/view/84
