Tech corporations promise that their new tools will make things easier.
AIs are the latest addition to nearly every product they ship, and Gemini’s integration into Gmail is just one example.
The idea sounds great. Tap a button to summarize a long email chain or add a few keywords and let the AI write for you.
Basically, let the machine do the heavy lifting.
But real life is usually a bit more complicated than that.
When the AI writes my email, and it reads it on your end, the message travels, but neither of us truly understands it, and that’s a problem.
I tried to let Gemini write my emails for a week — here is why I had to intervene
I let Gemini write my emails for a week and this happened
The cycle of AI fluff and AI summaries ruining conversations
What was meant to save us from drowning in text is flooding us with it.
Because it’s now basically free to generate professional-sounding emails, no one sends a quick “Yes” anymore.
People press the Polish button and turn a simple confirmation into an essay, which creates a ridiculous cycle.
The sender uses AI to fluff up a simple thought into a corporate masterpiece.
The recipient, overwhelmed by the wall of text, uses AI to summarize it back down to the original “Yes.”
The problem with this bot-to-bot loop is that it slowly wipes out all the human nuance in our communication.
Things like tone, hesitation, sarcasm, and what’s left unsaid are what AI can’t get right.
This may create a hallucinated agreement where everyone thinks they’re on the same page because the AI smoothed out all the rough edges.
The monotone culture of AI communication
Large language models (LLMs) are trained on the average of the entire internet. By definition, they regress to the mean.
As more people rely on them, business communication is being flattened into a beige monotone.
Different cultures really grow out of the way people talk to one another.
When we wash our communication through a “Polish this” filter, we strip away the spikes and edges that make a brand or a person unique.
As a result, emails are starting to all sound like they came from the same HR department. That’s why we are slowly developing a spidey sense for synthetic text.
Some words have basically become the giveaways of the LLM. If I see the words “delve,” “tapestry,” or “blend” in a casual Tuesday email, my guard goes up.
Real people don’t say, “Let’s delve into our Q3 marketing landscape.” Bots do.
When a human reads a message that feels slightly off — too polite, too structured, or stuffed with filler words — our trust filters close.
We dislike the writing and end up doubting the sender’s intelligence.
Believe it or not, a typo is starting to feel like a good sign. It proves someone was at the keyboard.
Writing to please AI first and people second
As the Summarize button becomes the default way to process info, our human writing is also changing to please the machine.
Similar to how bloggers use keywords to rank on Google, we’re doing Generative Engine Optimization (GEO) and writing with AI in mind first, and humans second.
I’m starting to notice people stuffing their messages with a “TL;DR” to make sure Gemini catches the main point.
There’s this fear that if a message is too nuanced or complex, the AI might mangle the summary or, worse, filter it out.
Bringing humans back into AI email writing the right way
Cutting out AI isn’t the answer to communication falling apart. But we need to add some friction back into the system.
I’ve put together a system. Now it might not solve everything, but it can definitely help with this problem.
Some emails should never be AI-generated
We have to be real. There are just some areas where tapping Generate is lazy and downright risky.
To keep ethics and trust intact, we need to draw a hard line in some areas.
For example, apologies and conflict resolution are strictly human territory. Automating an “I’m sorry” is an insult.
Keeping emails real with the drafter-not-sender approach
To stop the bot-to-bot loop, people need to be pulled back into the process.
One practical fix is implementing a “drafter, not sender” rule.
Gemini can help draft the structure of an email or document, but the human still has to rewrite the opening and closing lines themselves.
That small manual effort forces the brain to engage with what’s being said, rather than approving it on autopilot.
This will keep the message anchored in real human intent.
People are craving real, hand-crafted emails more than ever
As synthetic content floods our feeds, we’re starting to crave something real and unmistakably human.
Right now, receiving a genuinely hand-typed letter is becoming the ultimate status symbol.
It says, “I spent my finite life minutes on you.” That is a currency AI cannot counterfeit. And that’s something worth holding onto.
