Anthropic Wants Its AI Agent to Control Your Computer
Anthropic Wants Its AI Agent to Control Your Computer
Claude is the first major AI model to be able to take control of a computer to do useful work.
Y'all remember when we found out all the AI in the walkout Amazon grocery stores was just a bunch of dudes in an Indian call center watching cameras? That would be hilarious if this was the same, some random dude in India sshs into this dude computer then gets annoyed about it if date packages on the system and it spirals from there.
I see that Amazon Go nonsense parroted a lot here on Lemmy. It's not true.
Humans are involved in resolving issues (like self-driving cars that get stuck), and training, but it's not just a bunch of people watching cameras and updating a spreadsheet of your groceries.
The Verge has an article on the subject.
An impressive dead end.
Dude forgot he's using a tool that's not going to think for him
Boy am I glad all those new server farms are sucking up all the potable water...
Microsoft could have been ahead on this with Cortana if they had any real commitment and vision.
Their vision was to do screenshots every few seconds and start with that.
Someone made a wrapper for it you can run: https://github.com/corbt/agent.exe/
I'll pass.
No. But if anyone is interested in a opensourcer version of this check out open interpreter