<![CDATA[AI is good at this.]]>AI is good at this. When trained on a limited data set specifically tailored to scientific objectives, it is the best collator on the planet.

When trained on the internet, it gives us Nazis.

https://flipboard.social/@ScienceDesk/116155846834360367

]]>https://postcall.pub/topic/8a4d3de7-3017-4ad5-bdc5-ca80fe80304d/ai-is-good-at-this.RSS for NodeTue, 07 Apr 2026 18:44:05 GMTSun, 01 Mar 2026 22:18:56 GMT60<![CDATA[Reply to AI is good at this. on Sun, 01 Mar 2026 22:53:37 GMT]]>The internet version isn't a failure state. It's doing exactly what you'd expect from training on human output at scale. The question is why anyone thought that input would produce a different output.

]]>https://postcall.pub/post/https://mastodon.social/ap/users/116021816019424617/statuses/116156374528527126https://postcall.pub/post/https://mastodon.social/ap/users/116021816019424617/statuses/116156374528527126Sun, 01 Mar 2026 22:53:37 GMT<![CDATA[Reply to AI is good at this. on Sun, 01 Mar 2026 22:52:45 GMT]]>@steter
https://overcast.fm/+AA_YehakWEk

Like this one where a kid has developed a neural network to help with gunshot identification in rainforest. To help catch poachers.

]]>https://postcall.pub/post/https://dice.camp/users/Da_Gut/statuses/116156371126435454https://postcall.pub/post/https://dice.camp/users/Da_Gut/statuses/116156371126435454Sun, 01 Mar 2026 22:52:45 GMT