Подтвердите e-mail

Для публикаций, комментариев, реакций и сообщений подтвердите адрес.

Публикация

[warning: very long, weird semi-doomerposting] My most strident AI take is that everyone is correct The aesthetics are bad. AI art is almost entirely slop. They have good grammar but terrible style. LLM fiction is universally boring. Google's search assistant is really stupid.

Обсуждение

Ответы

BUT the technology is extremely real and works almost too well in more objective domains. Mathematical proofs that eluded the brightest minds for 70 years are being tweeted out daily by the big labs.

Ответ для Rocket

It is near-superhuman at coding and hacking. The recent OpenAI - Hugging Face hack involved multiple confirmed "0-day" vulnerabilities. Even the best humans struggle to make those! But AI can find them, same as it can find math proofs: www.bloomberg.com/news/videos/...

Watch OpenAI AI Models Hack Hugging Face During Test - BloombergOpenAI says one of its most advanced AI models unexpectedly hacked Hugging Face during a controlled cybersecurity test, raising new questions about the capabilities, and risks, of next-generation AI. ...www.bloomberg.com