Many everyday language features—from your phone’s autocorrect to the essays students draft with ChatGPT—trace back to the same breakthrough: let every word in a sentence pay attention to every other word at once. That trick, called self-attention, sits at the core of the transformer architecture unveiled in 2017, and natural-language research has shifted dramatically ever since.

Nearly three-in-ten U.S. adults have ever used a dating site or app, according to a 2023 Pew Research Center survey—a tipping point that signals romance’s migration into the swipe economy.

When the Federal Communications Commission issued a 13-page notice titled “Disclosure and Transparency of Artificial-Intelligence-Generated Content in Political Advertisements”, the agency took its first formal step toward forcing broadcasters to tell viewers whenever campaign spots rely on synthetic voices or imagery.

In our previous article, we explored how human motion data captured from VR devices is used to train robots. In this article, we will discuss the ethical risks associated with this harvesting of data from users, potentially without their consent.

In August 2023, researchers from the University of California, Berkeley told the USENIX Security symposium that they could pick the real person behind 55,541 publicly shared Beat Saber replays with 94 percent accuracy after just 100 seconds of head-and-hand motion. Lead author Vivek Nair called the result “unique and reliable identification,” according to USENIX.

In a recent episode of the "Reflections in Beige Podcast," Nathan A.M. speculated that motion data harvested from VR gaming devices could be used by companies for training robots. This speculation is not far from reality. Recent advancements in robotics have seen a significant shift towards leveraging human motion data to enhance robotic manipulation policies. This approach not only improves training efficiency but also enables robots to learn new motions directly from human demonstrations.

On a quiet January week in 2012, Facebook adjusted the emotional "lighting" for 689,003 unsuspecting users. By algorithmically subtracting about ten percent of either positive or negative posts from each person’s News Feed, the company set out to learn whether moods could spread without direct interaction. Two years later the peer-reviewed results landed in the Proceedings of the National Academy of Sciences, where researchers called the outcome “massive-scale emotional contagion.”

On a chilly Toronto night in October 2012, graduate student Alex Krizhevsky watched two consumer-grade NVIDIA GTX 580 cards hum on his bedroom floor. The GPUs slashed ImageNet training time and—more visibly—cut top-5 error from 26 percent to 15.3 percent, electrifying computer-vision research. What looked like a dorm-room hack would soon be cited by NVIDIA CEO Jensen Huang as proof that graphics hardware could fuel an AI renaissance.