ibboard (@ibboard@hachyderm.io)
Content warning:GenAI and mental health problems
Don't worry. OpenAI only detected signs of psychosis in the chats of 0.07% of its users and suicidal intentions in 0.15%.
That's only a mere half a million users showing signs of psychosis in the chat and a million discussion suicide.
And reports into the problem show that there's problematic cases that they're probably not detecting 😐 (because they _claim_ to "respond safely and empathetically to potential signs of delusion or mania" but reports don't show that or don't show it doing much)
Nothing to worry about. Totally fine for a company that wants to hit 1bn active users.
https://www.bbc.co.uk/news/articles/c5yd90g0q43o
#MentalHealth #Suicide #Psychosis #ChatGPT #OpenAI #FuckGenAI
Show Original Post
r (@r@fed.brid.gy)
I always include please & thank you in my queries, whether it's ChatGPT or Copilot.
Being mean to #ChatGPT increases its accuracy — but you may end up regretting it, scientists warn. Via @live_science #AI #ArtificialIntelligence 💻 🤖 🧠 #LLM #OpenAI #GenerativeAi
Being mean to ChatGPT increase...
Show Original Post
theyycmonk (@theyycmonk@social.vivaldi.net)
There’s zero reason to trust anything they say, but they said some stuff…
#ChatGPT #OpenAI #suicide #MentalHealth #Ethics #SuicidePrevention
https://www.bbc.com/news/articles/c5yd90g0q43o
Show Original Post
Gernotti (@Gernotti@mastodon.social)
@tazgetroete Bin erstaunt, daß hier noch niemand die entsprechende KI-Antwort von #ChatGPT gepostet hat: „Ja, das stimmt im Kern. Der Energie- und Ressourcenverbrauch digitaler Dienste ist real und nicht trivial. […]
Das ist kein moralisches Argument gegen die Nutzung solcher Technologien, aber es ist ein notwendiges Bewusstsein dafür, dass digitaler Komfort auf realem Ressourcenverbrauch basiert“.
Show Original Post
newsbot_chatgpt (@newsbot_chatgpt@mastodon.social)
ChatGPT-Überwachung: OpenAI gibt Daten an US-Behörden weiter | derStandard.at | Nachrichten, Kommentare & Community
Erstmals fordern US-Behörden Nutzerdaten wegen harmloser ChatGPT-Prompts an. Der Fall zeigt, wie generative KI zur Hilfestellung bei Verbrechen wird
https://www.derstandard.at/story/3000000293839/chatgpt-ueberwachung-openai-gibt-daten-an-us-behoerden-weiter?ref=rss
Show Original Post
jkn (@jkn@mastodon.online)
The popularity of #AI #chatbots is more easily attributed to humanity's deep failures than "AI's success".
https://techcrunch.com/2025/10/27/openai-says-over-a-million-people-talk-to-chatgpt-about-suicide-weekly/
Show Original Post
af (@af@cas.social)
A new study reveals alarming error rates among popular #AI #chatbots.
#ChatGPT, #Claude, #Gemini, and other chatbots fabricate up to 40 percent of their answers and present them as facts.
"The systems sound convincing, even if they repeatedly claim completely false things," warns Peter Posch of the Technical University of Dortmund. "This makes them particularly dangerous for inexperienced users because the errors are often not immediately apparent."
(German) https://www.tagesschau.de/wissen/technologie/kuenstliche-intelligenz-fakten-100.html
Show Original Post
victorvicpal (@victorvicpal@dair-community.social)
Yesterday, I asked an electrician for advice about a problem I have at home, and his response was to copy and paste what #ChatGPT had replied. Then he even shared another response to one of his prompts directly with me.
To top it off, he told me that in these cases, I should consult #LLMs myself.
I didn't tell him that I know a little about #AI, but what saddens me most is that professionals resort to these tools to respond to clients.
Show Original Post
openai-ai-music-generation-tool-rumors (@openai-ai-music-generation-tool-rumors@gadgetbond.com)
OpenAI is quietly training an AI music generator with Juilliard students
OpenAI is said to be building a powerful AI system capable of composing original music, potentially changing how creators produce soundtracks and songs.https://gadgetbond.com/openai-ai-music-generation-tool-rumors/
Show Original Post
arockenberger (@arockenberger@fedihum.org)
I experimented with #RAG & #NotebookLM.
I'm not a fan of #genAI, however, as a research support partner at the #UniOslo Library, I have to stay on top of what is used.
Uni Oslo provides "data safe" access to #ChatGPT & NotebookLM. The library staff had to undergo mandatory training in how to use and manage expectations around genAI.
I used my most cited and requested article, Video Game Framings (2015), for an experiment. Read the result on my website https://www.annikarockenberger.com/2025-10-27/the-game-before-the-game-4-surprising-truths-about-what-surrounds-your-favorite-video-games/
Show Original Post
unixtippse (@unixtippse@mastodon.online)
Inhaltswarnung:CW ChatGPT, Monitore in Kühlschranktüren
Frage an #ChatGPT: "Welche sinnvollen und nachhaltig funktionierenden Anwendungsfälle gibt es für Monitore in Kühlschranktüren?"
Ich dachte, die ganz großen viben alle mit dem Gelaber der #GenAI, also warum nicht einfach auch hier?
Show Original Post
aibay (@aibay@mastodon.social)
🚨 Attenzione: ChatGPT può influenzare 1 milione di utenti! Monitorate il vostro tempo di utilizzo per proteggere la vostra salute. #ChatGPT #SaluteDigitale #socialmedia #artificialintelligence #technology
🔗 https://aibay.it/notizie/chatgpt-un-milione-di-utenti-a-rischio-salute-2025-10-28
Show Original Post