human creators

Researchers at OpenAI and Microsoft have published a paper describing a 'doom loop' in which large language models, trained on data scraped largely without consent from human creators, degrade the open web and eventually poison their own training data. The paper reportedly acknowledges that people will come to see the models' wholesale ingestion of creative work as an unprecedented act of theft, reigniting debate over copyright and the sustainability of generative AI.

A Microsoft director has described AI training data scraping as 'the largest theft of labor in human history', while OpenAI's head of publisher strategy called ChatGPT an existential threat to publishers. The remarks were revealed in legal briefs filed in the New York Times' copyright lawsuit against OpenAI and Microsoft. The comments add to the debate over whether AI companies unfairly exploit journalists' and creators' work without permission or payment.

A new report claims OpenAI and Microsoft were aware that their AI products, trained partly on scraped web content, could trigger a 'doom loop' degrading the open web. According to The Verge, internal discussions suggested the companies understood that AI-generated content and content theft could crowd out human-made material, drawing comparisons to earlier warnings about Google's zero-click search results.

An artist posting in Spanish declares that artificial intelligence cannot create what they make by hand with markers and paint, nor emulate the unique feeling each person experiences while creating. They reject claims that AI is superior or inevitable, insisting it has not arrived to stay, and tag the work as human-made amid ongoing debates over AI art.