Sloppy literacy

It’s the time of year everyone tries to remember how to cook a turkey without creating something akin to sawdust.

It’s not just bang average poultry. The holiday season brings a surge of searches for recipes, as we all seek to indulge and/or impress.  

But this year, a new phenomenon is challenging recipe creators and leaving home cooks profoundly disappointed - AI slop. From impossibly perfect images of cookies to bastardised amalgams of trusty directions, AI is guiding many towards disaster - and a frantic Uber Eats order.

The issue isn’t just with recommending cakes are baked until the kitchen becomes an inferno, it’s that people don’t know how to spot AI-produced content. 

In one of the largest studies conducted, participants correctly identified AI-generated images 76% of the time, or “halfway between random guessing and perfect identification.” A smaller ABC experiment found high school students correctly identified AI content 67% of the time.

This is a problem that has far greater ramifications than doughy cookies. 

A recent ACMA study found 72% of Aussies believe they encountered some form of misinformation online. This probably implies 28% simply didn’t recognise it, rather than live in some wonderful internet bubble of universal truth.

A world where misinformation is everywhere, AI fakes are becoming more realistic by the day, and most people can’t decipher what’s genuine is creeping into dystopia.  

A natural response may be to look to suitable regulation. But with the social media age ban about to take effect and limit young people’s ability to improve digital literacy, and the government saying there is no need for AI legislation in Australia, that looks unlikely.

Instead, we are stuck with content we cannot trust and recipes that make no sense. Maybe ask for a cookbook for Christmas.