Even Meta's employees are struggling to grasp AI. Who could have predicted that?

      Unsplash

      If you're looking for an example of a tech giant pushing its employees towards an AI-driven future, Meta is a prime illustration. The company that built its success on extensive knowledge of its users has now turned that focus onto its own employees, leading to dissatisfaction among its workforce. Last month, Meta quietly notified tens of thousands of its U.S. employees that their corporate laptops would start tracking keystrokes, mouse movements, clicks, and screen activity. The intention was to incorporate that behavioral data into Meta’s AI models to understand how people actually interact with computers. The response was swift — within hours, internal discussion threads were inundated with expressions of anger, confusion, and emojis that clearly conveyed employee feelings.

      When an engineering manager inquired about opting out of the tracking, Meta’s chief technology officer, Andrew Bosworth, gave a straightforward reply: there was no option to opt out, at least not for company laptops. This is the same company linking AI tool usage to performance evaluations, conducting mandatory “AI Transformation Weeks” to retrain workers, and creating internal dashboards that gamify tracking employee consumption of AI tokens — a metric so rigorously monitored that some workers began creating AI agents to oversee their other AI agents. The entire situation began to resemble a self-consuming feedback loop.

      The layoffs only exacerbated the situation.

      This is not happening in isolation. On April 17, reports emerged that Meta intended to eliminate approximately 10% of its workforce — about 8,000 positions — with the first round of layoffs slated for May 20. Employees who had recently been encouraged to embrace AI, undergo training with AI, and now have their computer usage monitored to train AI were suddenly questioning whether they were, in fact, constructing their own replacements. To put it mildly, the timing was disastrous. Internal posts indicated that the atmosphere was “incredibly demoralizing.” At least three countdown websites were launched to track the days until the layoffs. Employees shared nihilistic memes, with one particularly notable post stating simply: “It does not matter.”

      Josh Edelson/Getty Images / Meta

      Mark Zuckerberg addressed the data collection issue during a company-wide meeting, presenting it not as surveillance but as a method to educate AI on how “smart people use computers to accomplish tasks.” He also remarked that AI represents “probably one of the most competitive fields in history” — a statement that resonated differently for employees sitting in an office, pondering their job security in just three weeks.

      This is merely an indication of what is to come everywhere.

      What is happening at Meta is not unique; it is just more advanced than what others are experiencing. Companies like Microsoft, Coinbase, and Block have made similar transitions recently, reorganizing around AI, which has resulted in layoffs and internal strife. However, Meta is executing all these changes concurrently and on a large scale: retraining employees, monitoring their behavior, linking job security to AI adoption metrics, and reducing headcount to support the entire initiative.

      Mariia Shalabaieva / Unsplash

      There is no straightforward way to manage any of this. An employee backlash over keystroke tracking in one of the world's most powerful tech companies — a company that has, among other things, been actively developing AI systems to observe and interpret human behavior — is a unique kind of irony. Meta spent years persuading billions to willingly share their data. Convincing its own employees to accept the same is proving to be far more challenging.

      Shimul is a contributor at Digital Trends, with over five years of experience in the tech space.

      Rice grain-sized sensor could give robots a delicate touch and prevent them from damaging items. Robots are highly accurate, but their gentle handling abilities can sometimes be lacking. A machine capable of assembling a car with almost flawless precision may still exert too much force in sensitive situations, such as within a human eye or during intricate surgery. This is why researchers at Shanghai Jiao Tong University are creating a novel type of force sensor that could enhance a robot's tactile feedback.

      The sensor, roughly the size of a grain of rice at only 1.7 millimeters wide, can be compact enough to fit inside advanced surgical instruments. What makes this sensor particularly intriguing is that it does not depend on conventional electronics. Instead, it utilizes light to assess force from all angles, including pressure, sliding movements, and twisting. Here's how it operates: at the end of an optical fiber is a soft material that slightly deforms upon contact with an object. This minute alteration affects the way light navigates through the sensor. The modified light pattern is transmitted through optical fibers to a camera, which captures it as an image. Researchers then apply a machine learning model to analyze the light patterns and convert them into accurate force measurements. In simple terms, the system learns to “interpret” touch using light alone, eliminating the need for an array of wires or multiple sensors within such a compact space.

      Read more

      Sci-fi got the gadgets right, but the vibes