Researchers warn about model collapse when LLMs train on AI-generated data

[Jan]

Model collapse — what happens when AI trains on AI-generated data

Researchers from the US and Canada have published a paper highlighting a significant problem facing the artificial intelligence (AI) industry — AI-generated content being used to train large language models.

While OpenAI's GPT, Meta's LLamA, and Google's LaMDA are currently honing their capabilities primarily on human-generated content on the Internet, the web could soon be filled to the brim with the content these models themselves create.

 

[DapperDanMan]

Soon your mother tongue will be unrecognisable to your ageing brain!

 

[JimboBob]

Maybe that's what the AI wants you to think, meanwhile it's planning world domination.

 

[gotaway]

Will this be the reason AI "helpers" turn against their makers? The rise of SkyNet?