There are 10 failures in Tech. I am going to introduce some failures related AI.
Woke AI blunder
GOOGLE GEMINI VIA x.com WOKENESS
People worry about bias creeping into AI. But what if you add bias on purpose? Thanks to Google, we know where that leads: Black Vikings and female popes.
Google’s Gemini AI image feature, launched last February, had been tuned to zealously showcase diversity, damn the history books. Ask Google for a picture of German soldiers from World War II, and it would create a Benetton ad in Wehrmacht uniforms.
Critics pounced and Google beat an embarrassed retreat. It paused Gemini’s ability to draw people and agreed its well-intentioned effort to be inclusive had “missed the mark.”
The free version of Gemini still won’t create images of people. But paid versions will. When we asked for an image of 12 CEOs of public biotech companies, the software produced a photographic-quality image of middle-aged white men. Less than ideal. But closer to the truth.
More: Is Google’s Gemini chatbot woke by accident, or by design? (The Economist), Gemini image generation got it wrong. We’ll do better. (Google)
AI slop
AUTHOR UNKNOWN VIA WIKIMEDIA COMMONS
Slop is the scraps and leftovers that pigs eat. “AI slop” is what you and I are increasingly consuming online now that people are flooding the internet with computer-generated text and pictures.
AI slop is “dubious,” says the New York Times, and “dadaist,” according to Wired. It’s frequently weird, like Shrimp Jesus (don’t ask if you don’t know), or deceptive, like the picture of a shivering girl in a rowboat, supposedly showing the US government’s poor response to Hurricane Helene.
AI slop is often entertaining. AI slop is usually a waste of your time. AI slop is not fact-checked. AI slop exists mostly to get clicks. AI slop is that blue-check account on X posting 10-part threads on how great AI is—threads that were written by AI.