AI content: genius or garbage? We ask AI

The first time anyone uses AI, it can seem mightily impressive. By providing a simple, natural-language prompt, you can get a sensible answer to seemingly any question. If you need to knock up a speech or an essay, or generate some copy for a website, services like ChatGPT, Claude or Grok can provide what looks like good-quality text in just seconds. If you need a striking image or video, AI services can provide that, too.

Yet after spending some time using these services, or perusing social media, another thing becomes clear: much of this output is ‘slop’, defined (by Google’s Gemini AI) as ‘low-quality, mass-produced digital content generated by artificial intelligence with minimal human effort’. Images of people with too many fingers or limbs. Award-winning novels that have been called out as AI-generated. LinkedIn profiles now hugely infected by computer-generated dross. As one Telegraph article noted, we are facing ‘a seemingly infinite wall of “slop” – easily-produced and generic AI-generated text that is ostensibly well worded but devoid of substance’.

Is all AI content really that bad? In this session, we’ll test the idea that such content is really without merit by asking three AI models to debate the merits of AI content. Do they actually have something to ‘say’? Are they comparable with human commentators? You be the judge.