Mastodawn

I have a friend who felt like he was between a rock and a hard place politically at work. So he wrote up the situation and fed it to an LLM to brainstorm for ideas.

It told him to quit. (In multiple related conversations.)

A mutual friend thought that using an LLM like that was a good idea and fed her situation into an LLM as well. It also told *her* to quit.

I would be interested in hearing more data points on this. 😆

#AI #LLM #chatbot #claude #chatgpt

Gert

6h ago

Vabbè, ora te lo dice pure la scienza: i #ChatBot sono concepiti e costruiti per darti sempre ragione. Ecco perché sono di una noia mortale. Meglio una rissa in un Rave Party 😂

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians
https://arxiv.org/abs/2602.19141

Sycophantic Chatbots Cause Delusional Spiraling, Even in Ideal Bayesians

"AI psychosis" or "delusional spiraling" is an emerging phenomenon where AI chatbot users find themselves dangerously confident in outlandish beliefs after extended chatbot conversations. This phenomenon is typically attributed to AI chatbots' well-documented bias towards validating users' claims, a property often called "sycophancy." In this paper, we probe the causal link between AI sycophancy and AI-induced psychosis through modeling and simulation. We propose a simple Bayesian model of a user conversing with a chatbot, and formalize notions of sycophancy and delusional spiraling in that model. We then show that in this model, even an idealized Bayes-rational user is vulnerable to delusional spiraling, and that sycophancy plays a causal role. Furthermore, this effect persists in the face of two candidate mitigations: preventing chatbots from hallucinating false claims, and informing users of the possibility of model sycophancy. We conclude by discussing the implications of these results for model developers and policymakers concerned with mitigating the problem of delusional spiraling.

arXiv.org

sayzard 11h ago

ITmedia AI＋ (@itm_aiplus)

교토부가 부립 고교생 약 1만 명에게 AI 영어회화를 도입했다는 소식이다. 외국인 방문이 많은 지역 특성을 반영한 교육용 AI 활용 사례로, 대규모 공교육 도입이 특징이다.

https://x.com/itm_aiplus/status/2039594955614855243

#ai #englishlearning #education #chatbot #kyoto

ITmedia AI＋ (@itm_aiplus) on X

京都府、府立高校の生徒約1万人に“AI英会話”導入　多くの外国人が訪れる都市で「英語は必須スキル」 https://t.co/8F7q0qXGXi

X (formerly Twitter)

Scripter

16h ago

Projekte in ChatGPT und Claude: So organisierst du wiederkehrende KI-Aufgaben | t3n
https://t3n.de/news/projekte-chatgpt-claude-erstellen-wiederkehrende-ki-aufgaben-1735861/ #KI #Chatbot #Anthropic #Claude #Projekte

Show thread

six_grandfathers_mountain 22h ago

@Theeo123 @adb
RE
I don't use #Lumo much...

Me too, but I tell others to check out the better #ChatGPT (I'm assuming, 🤣)

"Lumo Is the Only Chatbot That Protects Your Privacy"

"Most, if not all, AI #chatbots like #ChatGPT, #Gemini, (maybe NOT #Claude) use data collected from your conversations... if you're using the #chatbot for official work 🥶

#Lumo, has no incentive to farm your private data for ads AFAIK

https://www.makeuseof.com/proton-lumo-privacy-focused-ai-chatbot-test/

There Is Finally an AI Chatbot That Protects Your Privacy, and I've Tested It

Lumo is the privacy-focused AI chatbot we've been waiting for.

MakeUseOf

Daniel AJ Sokolov 23h ago

Great #AI hack! When you add a #chatbot to a website or app, make sure the chatbot bubble generally overlays important buttons or navigational elements.

Everybody else is doing it, so don't stay behind!