Skip to content
Hacker News · AI·· 1 h agoAI score38

Can We Fix AI? A Research Agenda for Ethically Trained LLMs

Can We Fix AI?

AI summary

A new essay series asks whether AI can be built without objectionable practices, starting with whether LLMs can be trained without "stolen data." The author argues AI code review is becoming essential for security-critical code, yet many people find the technology morally objectionable, and asks whether an ethically trained model could match Claude or ChatGPT at spotting security bugs. The first installment examines copyright and consent in LLM training data, flagging open research questions.

Source:Hacker News · AI · peterbloem.nl