AI is going to eat itself: Experiment shows people training bots are using bots

Re:AI is going to eat itself: Experiment shows people traini... by MagicMoonlight on Friday June 16, 2023 @11:55AM (#36358791)

That sums up about 99.9% of humans saying things.

So really there’s not a distinction between language models and people. Other than the lack of decision making and planning, the actual communication side is identical.

[ Parent | on HN ]

Re:AI is going to eat itself: Experiment shows people traini... by retrocryptid on Friday June 16, 2023 @12:23PM (#36359267)

I've seen no evidence ChatGPT has decision making or planning capabilities.

[ Parent | on HN ]

Re:AI is going to eat itself: Experiment shows people traini... by moffkalast on Friday June 16, 2023 @12:35PM (#36359499)

Then check out autogpt?

[ Parent | on HN ]

Re:AI is going to eat itself: Experiment shows people traini... by flangola7 on Friday June 16, 2023 @01:26PM (#36360464)

>No decision making ability

What prompts are you using? It helps me make decisions every day involving complex and nuanced criteria. GPT-4 (especially 32k API + plugins) is better than 95% of the directors, PMs, and CEOs I have worked under.

[ Parent | on HN ]

Re:AI is going to eat itself: Experiment shows people traini... by MagicMoonlight on Saturday June 17, 2023 @02:36AM (#36367884)

Exactly, that’s the feature we have that they’re missing

[ Parent | on HN ]

Re:AI is going to eat itself: Experiment shows people traini... by ramesh31 on Friday June 16, 2023 @12:15PM (#36359135)

I think thats what's so unnerving about this stuff and driving all the "yeah but it actually sucks" type of articles.

The possibility that we really are just stochastic parrots is too much for people.

[ Parent | on HN ]

Re:AI is going to eat itself: Experiment shows people traini... by retrocryptid on Friday June 16, 2023 @12:23PM (#36359282)

Language is a virus.

[ Parent | on HN ]