The AI crisis no one is talking about
03.04.2026 Оригинал: https://www.youtube.com/watch?v=ZcH5C8Jlltc Автор: Mo Bitar Asking ChatGPT about pi was the worst mistake he ever made. Become a member on YouTube: / @atmoio Book time with me: https://intro.co/MoBitar This video was inspired by this tweet: https://x.com/atmoio/status/2039426909440725158 Subscribe to Substack, where I share a written rendition of these videos: https://atmoio.substack.com Sources https://arxiv.org/pdf/2602.19141 https://techcrunch.com/2025/06/15/spiraling-with-chatgpt/ https://www.nytimes.com/2025/08/08/technology/ai-chatbots-delusions-chatgpt.html https://www.cbc.ca/news/canada/ai-psychosis-canada-1.7631925 https://torontolife.com/deep-dives/man-vs-machine-chatgpt-delusion-now-hes-suing-openai/ https://futurism.com/chatgpt-chabot-severe-delusions https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html https://www.thehumanlineproject.org/ --- LLMs were used on a very short leash to assist in the production of this video to: generate b-roll images and deep research. Narrative beginning, middle, and end, as well as color, analysis, conclusion, and most importantly the comic relief, are always written by me. Occasionally, I might keep elements from the LLM that I thought were useful to advance the narrative structure. It's important to note that I am not anti-AI, but rather anti being-stupid-with-AI (and stupidity abounds, which is what creates the opportunity to make these videos). One would be quite mistaken to treat this technology as completely useless. My consistent take on this channel has been that AI is supremely useful, but is not a tool that can automate or replace humans. It must be kept on a very tight leash. This limitation is not temporary, but rather inherit to LLMs. AGI would solve these problems, but is not something I'm convinced is possible.
03.04.2026 Оригинал: https://www.youtube.com/watch?v=ZcH5C8Jlltc Автор: Mo Bitar Asking ChatGPT about pi was the worst mistake he ever made. Become a member on YouTube: / @atmoio Book time with me: https://intro.co/MoBitar This video was inspired by this tweet: https://x.com/atmoio/status/2039426909440725158 Subscribe to Substack, where I share a written rendition of these videos: https://atmoio.substack.com Sources https://arxiv.org/pdf/2602.19141 https://techcrunch.com/2025/06/15/spiraling-with-chatgpt/ https://www.nytimes.com/2025/08/08/technology/ai-chatbots-delusions-chatgpt.html https://www.cbc.ca/news/canada/ai-psychosis-canada-1.7631925 https://torontolife.com/deep-dives/man-vs-machine-chatgpt-delusion-now-hes-suing-openai/ https://futurism.com/chatgpt-chabot-severe-delusions https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-chatbots-conspiracies.html https://www.thehumanlineproject.org/ --- LLMs were used on a very short leash to assist in the production of this video to: generate b-roll images and deep research. Narrative beginning, middle, and end, as well as color, analysis, conclusion, and most importantly the comic relief, are always written by me. Occasionally, I might keep elements from the LLM that I thought were useful to advance the narrative structure. It's important to note that I am not anti-AI, but rather anti being-stupid-with-AI (and stupidity abounds, which is what creates the opportunity to make these videos). One would be quite mistaken to treat this technology as completely useless. My consistent take on this channel has been that AI is supremely useful, but is not a tool that can automate or replace humans. It must be kept on a very tight leash. This limitation is not temporary, but rather inherit to LLMs. AGI would solve these problems, but is not something I'm convinced is possible.