Weāve got to start calling this stuff what it is: big autocorrect. I donāt want my health diagnosed by autocorrect. Or to watch a film written by autocorrect. Itās either a mechanical Turk, at best, or otherwise itās Clippy at the singularity. Itās not what theyāre saying.
Apple study exposes deep cracks in LLMsā āreasoningā capabilities āIrrelevant red herrings lead to "catastrophic" failure of logical inferenceā - Large Language Models (LLMs) are the computing technique behind ChatGPT, etc. ArsTechnica story by Kyle Orland: arstechnica.com/ai/2024/10/l...
Irrelevant red herrings lead to ācatastrophicā failure of logical inference.
AI alone means Google's GHG emissions have gone up by 50% since 2019. This is abhorrent, and in my view is a good enough reason to shut down all generative AI altogether. It's an obscene use of our remaining carbon budget. www.bbc.co.uk/news/article...
The tech giant says it will be difficult to meet its net zero target due to AI using so much energy.