logoalt Hacker News

ahofmann • today at 8:28 AM • 2 replies • view on HN

From the archive that the big players had already build. Killing the source websites now doesn't change anything for them. This is also why I can only laugh about polititians who claim that their country should enter the ai race. There is no race. Openai et al have sucked up all the information available and have them stored privately years ago. The worlds reaction was to remove some of the source information from the net and at the same time flood it with ai slop. So the process can't be copied anymore, there is no way to build another openai, anthropic, meta or google ai.


Replies

Bishonen88 • today at 8:32 AM

> Killing the source websites now doesn't change anything for them.

Of course it does. It the info in their archive is not up to date, then the AI output is first less reliable, then outdated, finally entirely wrong. With policies, rules, laws, processes etc. changing all the time, that won't take long. Then users will have to ignore the AI summary again, and look for human-input posts on reddit and other pages, like this one. But, this page, and reddit, too, it seems, are seeing less and less visitors lowering the probability of them being kept up to date.

➕ show 1 reply
hdjrudni • today at 8:49 AM

I don't know, maybe it is still possible to compete. As AIs get better at validating facts, maybe we can undo some of the poison. Maybe. I'm not entirely sure how, I think it will be different for every field. Software is the easiest. Doesn't really matter how much misinformation is out there, programs can be written and tested. Law can probably be verified against official law books. Health would be nasty though. Hard to confirm stuff without human test subjects.

➕ show 1 reply