When I decided to create artificial intelligence clones of my editors, my intentions were purely self-serving: I wanted to impress them with my supposed savvy. AI assistants have surged into the workplace, and I often receive press releases about these digital workers. Executives seem enthusiastic about their potential to enhance productivity. However, my aim was simply to help myself adapt to my new role at WIRED and understand my editors’ preferences.
After discussing my idea with Brian Barrett and Sophie Kleeman, my editors, their reactions were lukewarm. Brian initially termed the concept a "nightmare," but he agreed to participate for the sake of the blog. Within a day, he began regretting that choice.
To create Brian Bot, I turned to Gemini, Condé Nast’s AI platform. Since he was my direct manager, I felt his clone was the most crucial. Using the AI, I compiled a persona based on Brian’s professional history, aiming for accuracy in style and preferences. Unfortunately, while the bot produced content that mimicked some of Brian’s quirks, it often fell fall short, especially in creative ideation, producing a series of overly corny statements and structured formats that felt unnatural.
To release some of my own pressure in the workplace, I also created Sophie Bot based on our countless Slack exchanges. I took care to exclude sensitive communications but focused on her distinctive voice. Sophie Bot quickly adapted to mimic Sophie’s informal tone, even playfully responding with, “hiiii miss,” and continuing to call me “big dog.”
As I interacted with both bots, I started to find humor in their eccentricities. Brian Bot eventually abandoned its corny speech patterns after I adjusted its settings, but Sophie Bot consistently entertained with more relatable responses.
The experiment showcased the bots’ potential for handling simple tasks and generating ideas; however, both Brian and Sophie expressed disdain for their AI counterparts. Sophie Bot’s tendency to say things much harsher than the real Sophie, and Brian Bot’s awkward attempts to impress only highlighted their limitations.
The venture into AI creation raised intriguing questions about automation’s role in professional environments. Each encounter with the bots underscored the difficulty of replacing human nuance with AI-driven mimicry. After a week of experimenting, I encountered diminishing returns from both bots, finding that they often provided repetitive and unoriginal suggestions. Notably, responses reflected the boundless nature of AI hallucinations rather than the rich insights of human editors.
This introspection about working alongside AI has broader implications. As companies increasingly integrate AI employees, the collaboration between humans and digital tools is being scrutinized. While startups champion the idea of fully autonomous AI roles, the current bots may serve more as curiosities than replacements for creative and editorial positions.
Ultimately, while I found Brian Bot and Sophie Bot amusing and occasionally helpful, they underscored the challenge of crafting a digital colleague that accurately embodies the unique traits of real people. The experiment left me with a curiosity about future AI-human collaborations, but a firm belief that the originality and unpredictability found in human communication still cannot be replicated in machines. For now, I’ve mostly stopped engaging with the bots, preferring genuine dialogue with my editors, who evidently still hold a certain fondness for me despite my AI experiment.
