A WIRED writer recently built AI clones of his coworkers, feeding their chat logs and other data into a chatbot system. The results were less like polished digital assistants and more like exaggerated versions of the real people. One clone, Brian Bot, kept steering conversations toward improv comedy; another, Sophie Bot, repeatedly called the author "big dog." The quirks were not bugs but features of the underlying data.
The experiment is a single, anecdotal case, so it offers no broad evidence about how AI clones perform in real workplaces. But it does illustrate a practical point: when you automate a person's communication style, you also automate their tics, hobbies, and informal language. That can make the clones feel familiar, yet also unpredictable in ways that might unsettle colleagues.
For businesses eyeing AI "digital twins" of employees, the takeaway is caution. The technology can capture surface personality, but it may amplify oddities and miss context. The WIRED piece suggests that the future of work with AI clones will be less about efficiency and more about negotiating the strange, human-like behaviors these systems produce.