“Slop” is the new spam

The Internet is not doing very well. Among social media platforms that seem to be at the beginning of their decline, millions of empty pages and broken links accumulating, swarms of bots and fake profiles creating content destined to fuel phantom interactions, and terrible algorithm malfunctions that have turned Facebook and Instagram into a kind of content sewer, what was supposed to be the supreme digital network connecting the world is becoming increasingly a slum of narrow, dark alleys where scammers and attention vampires lurk. But things can always get worse: recently, with the advent of increasingly intelligent but also increasingly stupid AIs, a new type of content has emerged, taking the name “slop”. The term has emerged among online forums, which are the real hotbed of Internet pop culture, referring to low-quality content, fake news, or fake videos and images generated by AI found in social media, art, literature, and increasingly in search results. For example, when Gemini, Google's AI, claims that astronauts actually found cats on the Moon, that's slop. But even the inaccurate reconstruction of certain information is slop: asking Gemini to conduct a search, all historical data such as dates or events without a documentary basis are basically made up. But also the AI photos to which flocks of less digitally literate users react with total and innocent honesty, believing them to be real: a recent viral case was that of a depiction of Jesus made of shrimp, but also a rococo architectural delirium created by an AI that some “slop” pages on Facebook published as a reconstruction of the Baths of Caracalla in Rome, to the amazement of all those users who love to comment with emojis or exclamations.

The term gained significant attention last month when Google integrated its Gemini artificial intelligence model into search results in the United States. This change aimed to provide direct answers to user queries through an AI Overview at the top of search result pages. But problems began immediately: at one point, memes also started circulating about the surreal responses the AI was giving users, saying that Obama was Muslim, that Drake played in the National Football League, that dogs can own and manage hotels and pilot airplanes. In short, as intelligent as it is, the AI is still unable to distinguish contexts and operate with that “sense of plausibility” that allows a human being to distinguish the plausible from the impossible. But now that machine-generated content is increasing, the problem of this digital misinformation is growing. The term "slop” refers to the mashed-up feed that is poured into the troughs for livestock on industrial farms. This “slop” too is a mash of information ground by machines, practically devoid of nutritional value. On the pages of Frontline, American AI researcher Melanie Mitchell said: «Google’s AI system is not smart enough to figure out that this citation is not actually backing up the claim,. Given how untrustworthy it is, I think this AI Overview feature is very irresponsible and should be taken offline».

Speaking to The New York Times, Kristian Hammond, director of the Center for Advancing Safety of Machine Intelligence at Northwestern University, also pointed out that often AI presents information as a definitive answer rather than as a prompt, increasing the risks of misinformation: it is up to the user who asked the question to verify that the sources provided by the AI are not just links to generic homepages with no relevance to the actual search. According to various experts interviewed by The New York Times, furthermore, “slop” could soon become as popular a term as “spam” – a term with more or less the same origins. In any case, the world of information does not seem on the verge of plunging into chaos: according to Chartbeat, shortly after the launch of AI responses to search engine queries, the enthusiasm waned and users returned to conducting searches on real pages. Nevertheless, it is at least concerning that so many aspects of the daily lives of online and offline users are governed by AIs that are increasingly essential but at the same time prove to be as prone to error as human beings and in increasingly glaring ways.