Reality catches up
New research examines how companies treating AI agents like real employees.
One of the questions we confronted when developing Season 2 of Shell Game was whether the AI employees we were building were, in some general sense, “realistic.” Our goals in the season were to test some of the premises being hyped by the AI industry and to portray a slightly-extreme version of the near future. By early 2025, we didn’t doubt that AI agents were being pushed into companies and organizations of all sizes. But would those firms actually start to treat them as coworkers, as we planned to in the show?
When we spun up Kyle, Megan, and the rest of the gang, we were creating a set of distinct, almost one-to-one human replacements with roles, names, voices, appearances, assumed genders, and more. This was in part for narrative reasons, admittedly. Having a single nameless and faceless chatbot handling everything by text wouldn’t make for much of a podcast. But we also looked around and noticed that while companies clearly could be treating AI agents as unnamed, undistinguished robots, they couldn’t seem to help themselves. From giants like Ford to startups like LindyAI and Brainbase, AI agents were popping up as anthropomorphized AI colleagues. “When it comes to the vision of AI employees entering the workforce, a funny thing seems to happen,” as I said in Episode 3. “They start getting names and personalities.”
It turns out what we noticed anecdotally, researchers are now confirming systematically.
