Photo by Conny Schneider on Unsplash

Last Tuesday, I received an email from a colleague that made my skin crawl. It wasn't offensive or inappropriate—quite the opposite. It was perfectly polite, grammatically flawless, and completely devoid of personality. Three sentences in, I knew ChatGPT had written it. Not because it was bad, but because it was aggressively, unnaturally perfect.

We're living through a strange moment in professional communication. Generative AI has become powerful enough to handle complex coding tasks, analyze medical imaging, and write coherent reports. Yet something fundamental breaks when we ask these systems to do something humans have been doing for centuries: write an email that sounds like a real person.

The irony cuts deep. We built AI systems by training them on millions of human texts, and now those same systems are struggling to capture what makes human communication actually work. That's worth examining.

The Uncanny Valley of Corporate Speak

ChatGPT doesn't understand context the way humans do. When you ask it to write a professional email, it defaults to a kind of sterile formality that hasn't actually existed in workplaces since about 1987. You know the voice: the one where every sentence feels like it was workshopped by a committee of HR departments.

"I trust this message finds you well." No. Nobody actually says this. People say "Hey," or "Hi," or they skip the greeting entirely and get straight to business. But ChatGPT, trained on billions of emails and corporate correspondence, learned that formal = professional. It missed the actual evolution of workplace communication over the past twenty years.

Consider a real example. When a human writes "Can we sync up tomorrow?" they're conveying casual urgency, availability, and a desire for quick resolution. When ChatGPT writes the same phrase, it sounds like a prompt from a flowchart. The words are identical, but something essential has evaporated.

This matters because email is where work actually happens. It's not technical writing. It's relationship building, negotiation, and persuasion wrapped in casual language. The moment your email reads like it was generated by an algorithm, you've lost credibility before anyone even reads past the opening line.

The Problem With Consistency

Here's a paradox: AI systems are too consistent. Humans contradict themselves constantly. We use different tones for different people. We might be formal with our boss but absurdly casual with our teammates. We use inside jokes, reference previous conversations, and occasionally make grammatical choices that would horrify an English teacher but perfectly capture what we mean.

ChatGPT can't do any of that effectively. When you ask it to write an email, it produces something that sounds the same whether you're writing to your CEO or your friend in accounting. It doesn't know who you actually are, what you actually care about, or how you actually talk.

More problematic: it doesn't know what you've already said. One of the most important aspects of professional communication is consistency over time. Your colleagues know how you write. They know you're the person who signs emails with "—Sarah" rather than "Best regards." They know you end sentences with "thoughts?" instead of "I would appreciate your thoughts on this matter." They know your voice.

When you suddenly start using ChatGPT for emails, you're creating a version of yourself that your colleagues don't recognize. Some people won't consciously notice it, but they'll feel it. Studies on digital communication show that people detect authenticity violations faster than we'd expect—usually within the first thirty seconds of reading something.

The Emotional Intelligence Chasm

Email is often where we navigate emotional complexity at work. Someone's frustrated but trying not to show it. You need to deliver bad news while maintaining goodwill. A colleague is clearly upset about something and you need to acknowledge it without making things worse.

This is where AI fundamentally fails. ChatGPT doesn't understand emotions; it predicts statistically likely patterns. Feed it a prompt like "Write an email declining someone's request without hurting their feelings," and it will generate something technically appropriate but emotionally tone-deaf. It'll be too apologetic or not apologetic enough. It'll miss the social cues that tell you how much reassurance is actually needed.

Consider this scenario: A junior employee asks for feedback on a project that's actually quite weak. A human manager might write something like, "Good start here—I see what you were going for. Let's grab coffee and talk through some directions that might work better." It's honest but gentle, specific but forward-looking.

ChatGPT would produce something longer, more formally structured, and ultimately less reassuring. It might include phrases like "constructive feedback" or "areas for improvement" that technically communicate the same information but create anxiety instead of safety.

And that anxiety? It gets remembered. It shapes future interactions. It's the difference between maintaining a functional working relationship and slowly eroding trust.

When AI Reveals What We Actually Value

The failure of ChatGPT at workplace email is actually revealing something important: we care deeply about authenticity in professional settings. We care about people. We want to know we're dealing with actual humans, not algorithmic approximations of them.

This becomes even more important when you consider the broader surveillance and privacy questions at play—as many companies are already tracking communication patterns in ways that make artificial voices even more unsettling.

The best emails I receive are the ones that feel like they came from a specific person on a specific day, dealing with a specific situation. They might have typos. They might be slightly rambling. They might reference something I mentioned in person last week. They feel real because they are.

Right now, we're in a phase where most people use ChatGPT for emails occasionally, and when they do, other people notice. But what happens if this becomes ubiquitous? What if everyone's email sounds like a corporate template? We'll lose something valuable—the texture of actual human connection.

For now, your best move is probably to keep using your own voice. Your colleagues will thank you. More importantly, you'll still sound like yourself.