@SigaGazetaBR AI é tecnologia incrível, mas considerem que a inteligência real é human in the loop, por isso chamemos de artificial inference. Não vamos promover a inteligence, mais correto inference.
@elonmusk put them on space datacenters, add unlimited gpu, internet and power. then we can try Matrix solution: burn the sun itself to stop machines. lol
@ZixuanLi_@huggingface@OpenAI@Zai_org that's why there is another concern here... models and data attested and distributed, independent of regulations and that kind of asymmetric conditions.
@ClementDelangue@OpenAI@ClementDelangue, can hf provide methodology about the evidence you cite that models, datasets, and metadata were not tampered? hf data correctness is fundamental to our community safety and trust here is crucial.
Frontier models make mistakes 10-20% of the time. Effects are real, and there are almost no scientific studies on how to handle it. We need to review and fix. The more they produce, the more we need to fix.
Effects are real and the industry needs to act.
I'm starting to write my own code again and after that ask to agent fix. Tired fixing their errors all day.
The burnout is real and nobody is talking about it.