AI Agents as "Coworkers" May Impair Human Performance

Treating artificial intelligence agents as "coworkers" can negatively impact human performance, according to research conducted by Boston University professor Emma Wiles. In a study, managers who were led to believe their work was being done by an "AI employee" named Alex, complete with a title and responsibilities, made 18% more errors compared to those who attributed the work to a chatbot. This finding suggests that the increasing trend of companies like Microsoft, OpenAI, Anthropic, and Google advertising AI agent teams as "digital colleagues" may be a flawed approach for worker integration.

The research highlights a potential pitfall in the development and deployment of advanced AI systems. As AI agents become more sophisticated and integrated into workflows, the way humans perceive and interact with them could significantly influence productivity and accuracy. The study's results indicate that anthropomorphizing AI tools to the point of treating them as human-like colleagues might inadvertently undermine the very efficiency gains these technologies are intended to provide. This perspective challenges the current trajectory of Silicon Valley in promoting AI agents as direct replacements or equivalents to human team members.

Separately, the company Sceye is preparing to test a stratospheric platform designed to deliver enhanced internet connectivity. This approximately 200-foot-long, solar-powered craft is scheduled to launch as early as August. Its mission involves positioning itself in the stratosphere, about 18 kilometers above the ocean's surface, over the southwestern US. From this altitude, it will utilize a custom antenna to augment existing 5G networks by beaming data directly to devices. This test aims to demonstrate the capability of High-Altitude Platform Stations (HAPS) to provide internet services from the upper atmosphere.