We’ve been on the threshold of an abundance of mass creation of AI generated content, both human and machine generated. As discussed last month, it’s now called ‘AI Slop’ with increasing frequency in the media. It’s something we’ve been ‘on the verge’ of for some time now, with concern about AI generated ‘deepfakes’ to existential questions on ‘What is a Photo’, to coming waves of both ‘Synthetic Data’ now generating torrents of ‘Synthetic Content'. OpenAI signing up over a million users in a day with users generating ‘Ghibli’ style images recently was a vivid case in point.
Not to mention increasing use of ‘Digital Twins’ for both enterprise and consumer use. And the increasing generation of ‘machine to machine’ (m2m) content for ultimate mainstream consumption. Turbo charged by AI Agents soon, no less. 
This issue has come to the forefront again with Google unveiling its answer to OpenAI’s text to video AI Sora models with Veo 3 at last week’s Google I/O conference which had over two hundred mentions of ‘Gemini AI’ both as one phrase and two words. It’s new trick is the ability to generate realistic audio and dialogue to AI generated video. So far the text to video had been like ‘silent films’ of yore. With rising concerns about the implications for Hollywood.
And even though Veo 3 in a usable form is behind a hefty paywall of $250/month subscription, the internet has been flooded with examples of creative folks can do with it, both good and bad. Fortunately guardrails have blocked off the really bad stuff, but the waves of concern are rising in the media. 
The Verge lays it out in “Google’s Veo 3 AI video generator is a slop monger’s dream”:
“Even at first glance, there’s something off about the body on the street. The white sheet it’s under is a little too clean, and the officers’ movements are totally devoid of purpose. “We need to clear the street,” one of them says with a firm hand gesture, though her lips don’t move. It’s AI, alright. But here’s the kicker: my prompt didn’t include any dialogue.”
Veo 3, Google’s new AI video generation model, added that line all on its own. Over the past 24 hours I’ve created a dozen clips depicting news reports, disasters, and goofy cartoon cats with convincing audio — some of which the model invented all on its own. It’s more than a little creepy and way more sophisticated than I had imagined. And while I don’t think it’s going to propel us to a misinformation doomsday just yet, Veo 3 strikes me as an absolute AI slop machine.
Journalists and online creators are having a blast through the Memorial Day weekend creating all kinds of examples, both cool and clever, AND crass and cringeworthy. Mirroring human nature as it were. Just with a new technology toy. A toy for now, a useful tool for millions soon enough.
“Google introduced Veo 3 at I/O this week, highlighting its most important new capability: generating sound to go with your AI video. “We’re entering a new era of creation,” Google’s VP of Gemini, Josh Woodward, explained in the keynote, calling it “incredibly realistic.” I wasn’t completely sold, but then, a few days later, I had Veo 3 generate a video of a news anchor announcing a fire at the Space Needle. All it took was a basic text prompt, a few minutes, and an expensive subscription to Google’s AI Ultra plan. And you know what? Woodward wasn’t exaggerating. It’s realistic as hell.”
