Background: Deep learning has gotten pretty good at generating text descriptions for images. So good, in fact, that researchers decided to take it a step further with a neural network that can create free verse using an image as the source of inspiration.
How it works: To learn how to use an image as its muse, an algorithm was trained on thousands of image/poem pairs. That algorithm was then teamed up with a generative adversarial network (GAN) to look at images and produce poems from scratch.
Results: In a kind of poetry Turing test with over 500 human judges (including 30 English majors who were considered “experts”), AI-created verse was evaluated against poems written by people. Let’s be honest here: the AI “poetry”(see above) reads like something by a particularly angsty teenager. But the English majors were only able to spot the automated poem about 60 percent of the time. The rest of the evaluators weren’t any better than chance at picking out the AI.
A Roomba recorded a woman on the toilet. How did screenshots end up on Facebook?
Robot vacuum companies say your images are safe, but a sprawling global supply chain for data from our devices creates risk.
The viral AI avatar app Lensa undressed me—without my consent
My avatars were cartoonishly pornified, while my male colleagues got to be astronauts, explorers, and inventors.
Roomba testers feel misled after intimate images ended up on Facebook
An MIT Technology Review investigation recently revealed how images of a minor and a tester on the toilet ended up on social media. iRobot said it had consent to collect this kind of data from inside homes—but participants say otherwise.
How to spot AI-generated text
The internet is increasingly awash with text written by AI software. We need new tools to detect it.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.