TL;DR
A reporter handed his inbox to Google Gemini for five days. The AI sped through routine tasks at first but soon produced polished, repetitive language that dulled personal tone, prompting him to limit the tool to drafting and synthesis rather than final send-off.
What happened
Ben Khalesi ran a five-day experiment letting Google Gemini generate his email replies inside Google Workspace. For the first three days Gemini handled mundane tasks—scheduling, confirmations and quick refusals—efficiently by using thread context and, when Workspace access was enabled, checking Google Drive and Calendar. A cited example: the reporter asked Gemini to find a March receipt for Dell monitors in Drive and draft a reply; the model completed the task end to end. By day four the output started to sound uniformly polished and filler-heavy, repeating lines like “Hope this finds you well” and other neutral openings that colleagues noticed as insincere. On day five Khalesi relegated Gemini to “junior intern” work: structuring content and summarizing threads, but not sending sensitive messages. He established ground rules—always edit AI drafts, don’t use AI for apologies or negotiations, and verify summarized action items—and reported feeling mentally rusty when he returned to writing messages himself.
Why it matters
- AI can substantially reduce time spent on repetitive email tasks but may erode personal tone in ongoing conversations.
- Automated replies that rely on neutral, filler language can make communication feel insincere and affect workplace relationships.
- Users need guardrails and verification steps because AI can miss emotional nuance and negotiation subtleties.
- Relying too heavily on AI risks reducing users’ readiness to produce original messages and may cause short-term writing fatigue.
Key facts
- The experiment lasted five days and was conducted by Ben Khalesi for Android Police.
- Gemini performed strongly on routine, coordination tasks during days one through three.
- With Workspace enabled, Gemini can cross-check information in Google Drive and Calendar to inform replies.
- An example task: locating a March receipt for Dell monitors in Drive and drafting the response, which Gemini handled.
- By day four the AI’s responses adopted repetitive, polished phrasing that colleagues found lifeless.
- By day five the reporter used Gemini for drafting and synthesis but edited outputs before sending.
- Ground rules formed by the reporter: never send without editing, avoid AI for apologies and negotiations, and use it for thread summaries with verification.
- After heavy use of AI, the reporter experienced difficulty writing a simple email unaided.
What to watch next
- Whether Gemini’s tone controls or personalization features improve to better preserve individual voice — not confirmed in the source.
- How organizations set policies for AI use in sensitive communications such as apologies and negotiations — not confirmed in the source.
- Longer-term effects on users’ independent writing skills and workplace communication norms — not confirmed in the source.
Quick glossary
- Gemini: A large language model integrated into Google Workspace that can draft and generate text based on context.
- Google Workspace: A suite of cloud productivity tools from Google that includes Gmail, Drive, Calendar and other collaboration services.
- Prompt: A user-provided instruction or request that guides an AI model’s response.
- Thread synthesis: Summarizing the key points and action items from an email conversation.
Reader FAQ
Did Gemini save time in the experiment?
Yes — the reporter found it sped up scheduling, confirmations and simple coordination tasks during the first three days.
Did Gemini preserve the reporter’s personal voice?
No — the AI produced polished, repetitive phrasing that was perceived as lifeless and insincere.
Is it safe to send AI-written emails without editing?
The reporter’s rule was: always insert, edit, then send; he advised against sending unedited AI outputs.
Can Gemini access my Drive and Calendar automatically?
When the Workspace setting is enabled, Gemini can cross-check Drive and Calendar for context, per the report.
I tried to let Gemini write my emails for a week — here is why I had to intervene Credit: Lucas Gouveia / Android Police By Ben Khalesi Published 31 minutes…
Sources
- I tried to let Gemini write my emails for a week — here is why I had to intervene
- I Tried Gemini For My Email. Here's Why I Don't Trust It
- Google's Gemini AI Takes Email Drafting to the Next Level
- A guide to Gmail draft refinement: Using AI to write better …
Related posts
- Leaked preview shows Samsung integrating Perplexity into Bixby on Galaxy S26
- Safe CEO Don Murray: AI should assist human experts, not replace them
- AI Futures Model December 2025 Update: New Timelines and Takeoff Forecasts