A Wired reporter created AI clones of his editors using Gemini and encountered the limits of AI personalization
A Wired reporter used Gemini to create AI clones of two of his editors from their articles, podcasts and Slack messages. The clones captured the basic communication style but slipped into typical AI clichés and, on one occasion, drew on an incorrect source about another person.
A reporter for Wired who covers the future of work used Gemini – the official AI platform at Condé Nast – to create AI clones of two of his editors, Brian Barrett and Sophie Kleeman. The aim was to better understand their communication styles and be a better colleague. Gemini analyzed the publicly available digital footprint of Brian Barrett (articles, podcast transcripts, PR materials) and generated documents titled “Writing and Editing Style Guide”, “Persona and Style Guide” and “Editorial Dossier”.
According to the article, the so-called Brian Bot was not good at helping with article ideas, and its responses contained typical AI elements – bold subheadings and bulleted lists. When preparing the profile, Gemini initially also drew on PR material belonging to another person with the same name, which the author subsequently removed from the source materials. The bot did, however, make a claim about a “yes, and” leadership style derived from Brian's past experience with the improv theater Upright Citizens Brigade – the reporter subsequently checked this specific fact with Brian and confirmed it was true, even though he himself had not previously spoken about it.
The second clone, Sophie Bot, was created from approximately a month's worth of Slack conversations (around 17 000 words used to create the source materials) and, according to the article, captured the tone and communication style of Sophie Kleeman significantly better than Brian Bot. The text also mentions an anonymized employee at a large technology company identified as “Chip”, who uploads all emails, chats and documents from his manager to a Gemini Gem. He says the tool helps him with proofreading, debugging code and preparing for meetings, making him more efficient.
The available article text ends midway through the description of Sophie Bot, so the final assessment of the experiment is not available. You can find details in the source article.
Why it matters
The experiment specifically shows where AI clones of real people fail – they generate generic AI phrases instead of the actual style of the person and may pull facts from an incorrect source. For individuals, this means having to verify what an AI persona claims about colleagues; for companies, it raises the question of how to regulate the uploading of internal communications to AI tools that employees are already using informally.
Two audiences, two different impacts
What this means
For individuals
The experiment shows that an AI clone of a colleague can sound credible while inventing or misattributing facts about that person – content from similar personas therefore needs to be verified, rather than blindly trusted.
For a business
According to the article, employees are already informally creating AI copies of their managers and feeding them company communications without a clear company policy in place – this poses a risk to privacy and company data.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.