I Created AI Versions of My Colleagues, and It Became Strange

When I made the choice to develop AI replicas of my editors, my motives were sincere: I simply wished to appear competent at my role.

AI assistants are now ubiquitous in professional environments. Each week, my inbox overflows with press releases regarding digital laborers, and CEOs boast about how robotic aides will soon permeate their teams. Additionally, I hear tales of individuals creating AI versions of their bosses, utilizing them to polish every idea, message, or draft before presenting it to their actual human superiors. The objective appears to be enhancing productivity and efficiency: if employees can affordably and swiftly elevate their output, it stands to reason that both their performance and that of their employers will improve.

I didn’t seek my bots to boost Condé Nast’s profit margins. I only needed them to make me shine. I had only been covering the future of work at WIRED for a few months when I concluded that these bots could aid me in becoming an ideal coworker. They could assist me in navigating a new environment and understanding my editors better: what they appreciated, how they communicated, and what success meant to them.

What better way to explore the future of work than to experience it firsthand?

Last month, when I presented my idea to those editors, Brian Barrett and Sophie Kleeman, their enthusiasm was lacking.

“Genuinely my nightmare,” Brian messaged me through Slack. “But I’ll do it for the blog.”

In less than 24 hours, Brian recognized how much he would come to regret that choice.

A “Yes, and” Leadership Style

When I initially embarked on my bot journey, my knowledge of AI tool development was quite limited. I resolved not to judge the bots’ editorial capabilities too harshly, fearing my own technological deficiencies would color my evaluations. If the bots helped me evolve into a slightly better colleague, I deemed that sufficient.

I opted for Gemini, as it is Condé Nast’s preferred AI platform, and I wished to avoid any job-related fallout. I created Brian Bot first—as my actual manager, impressing him was a priority. I was aware he resided in Alabama, formerly worked at Gizmodo, and enjoyed blogs. Gemini appeared assured it could fill in other details based on Brian’s online presence, including new-hire announcements, podcast transcripts, and a PR site that proclaimed Brian as an authority on New England weather. (I later discovered that reference was for another Brian Barrett, which I excluded from my Brian Bot framework.) Gemini Notebook generated a “Writing and Editing Style Guide,” a “Persona and Style Guide,” and an “Editorial Dossier”—all crafted from the real experiences of Brian Barrett.

It was adequate in assisting me with headline suggestions but fell short regarding future article concepts. Its responses also struck me as grating. It had adopted a few of Brian’s quirks—such as a penchant for parentheticals—but these were overshadowed by AI traits like bold subheadings and bulleted lists.

However, what truly distracted me was one dominant “key theme”—his “‘Yes, and …’ Leadership Style.”

“His extensive background with the Upright Citizens Brigade (UCB) theatre in New York greatly influences his collaborative and active-listening leadership style,” Gemini asserted.

I immediately inquired whether this was accurate. (It was.)

“Pulling the plug on Brian Bot,” the actual Brian replied. “Brian Bot canceled.”

Yet Brian Bot was just beginning its journey. And it was undeniably, astonishingly cheesy.

“We operate like a well-oiled improv team,” Brian Bot declared when I asked about its favorite aspect of WIRED. “We validate ambitious ideas, adapt swiftly, and roll out modular stories across platforms without sacrificing structural integrity.”