#AIcannotBeTrusted
I talked to the AP about grok's "white genocide" meltdown. A few hours after publication, xAI admitted to hard coding it. This is important accountability journalism because people use AI as an arbiter of truth & here we see AI owners making the tools match their politics
apnews.com/article/elon...
Why was Elon Musk's AI chatbot Grok preoccupied with South Africa's racial politics?
Much like its creator, Elon Musk’s artificial intelligence chatbot Grok was preoccupied with South African racial politics on social media this week, posting unsolicited claims about the persecution a...
apnews.com
May 16, 2025 at 12:50 PM
The Trump FDA rushed to integrate half-cooked "AI" that "struggles with simple tasks" and often provides "inaccurate or only partially-accurate" answers to basic questions about FDA-approved products and other industry issues

are we perhaps detecting a theme
FDA’s AI tool for medical devices struggles with simple tasks
The agency announced Monday that a separate AI tool, called Elsa, had been rolled out to all FDA employees. Sources say it also has issues.
www.nbcnews.com
June 5, 2025 at 1:59 PM
Interesting to see OpenAI adopt the reporting style of major platforms for state-sponsored campaigns.
NPR - Chinese propagandists are using ChatGPT to write posts and comments on social media sites — and also to create performance reviews detailing that work for their bosses, according to OpenAI researchers. https://www.npr.org/2025/06/05/nx-s1-5423607/openai-china-influence-operations
June 5, 2025 at 1:17 PM
When you peer inside a large language model’s black box, you can see what it tells you when it thinks you’re rich—and what it tells you when it thinks you’re not.

@zittrain.bsky.social on the assumptions AI makes about people:
What AI Thinks It Knows About You
What happens when people can see what assumptions a large language model is making about them?
bit.ly
May 21, 2025 at 1:17 PM
#AIMakesBigErrors reason I am posting is because I just solved a tricky equation. Yesterday working a problem in Calculus, had AI give me some help. In it the genius AI corrected my assumption that dx/dx is 1, saying ' that is wrong'. Several Textbooks here say it isn't. #AIcannotbetrusted!
November 6, 2025 at 6:28 AM