BBCNEWS Asia Business Report June 4, 2024 01:38:00
About the risk of malicious actors deliberately spreading disinformation in the near—term, i actually think it's a bigger danger that people are starting to use these models as though they're search engines, even though it's well known that they oftentimes make up facts and claim things that aren't actually true and i thinkjust that might actually cause lots of misinformation, potentially some meaningful confusion in the future. the copyright issue is potentially even harder. the way that these models work is that after they're trained on some data, they have a sort of inherent propensity to...