Should You Let AI Write Your Emails?

Your email client offers to write your reply. It has analyzed the incoming message, understood the context, and generated three paragraphs that express exactly what you would have said. You read it, nod, click send. The recipient will never know. Should they?

This question is arriving in millions of inboxes as AI writing assistants become standard features. And the answer is not as obvious as it first appears. Communication is not just about information transfer. It's about relationship, intention, and the peculiar human practice of encoding thought into language.

When someone writes to you, they are giving you something. Their time, their attention, the effort of translation from thought to word. When you reply with AI, what are you giving back?

The Case for AI Assistance

Let's start with the obvious benefits. Email is often tedious. Most messages don't require deep personal engagement. Scheduling confirmations, routine updates, acknowledgments of receipt. These communications follow predictable patterns and serve primarily logistical functions. Having AI handle them frees you for work that matters more.

There's also the question of accessibility. For people who struggle with writing due to disability, dyslexia, or language barriers, AI assistance is transformative. It allows them to participate in written communication at a level that might otherwise be impossible. Dismissing AI-assisted writing entirely means dismissing these users.

And we already accept many forms of writing assistance. Spell check, grammar suggestions, templates, ghostwriters for executives. The line between "your writing" and "assisted writing" has always been blurry. AI just makes the assistance more powerful and more available.

Tell me more about outsourcing cognitive tasks

The Case Against

But something is lost. When you write to someone, you engage in an act of attention. You think about them, what they said, how to respond. This engagement is part of what makes communication valuable. It's not just the words; it's the fact that someone took time to construct them.

AI-generated replies look like this engagement but aren't. They simulate attention without requiring it. The recipient believes they received something personal when they received something generated. This is a form of deception, even if no explicit lie was told.

There's also the question of what happens to your own thinking. Writing is not just output; it's a form of processing. When you compose a message, you clarify your thoughts, work through your reactions, understand the situation better. Outsource the writing and you outsource the thinking.

The cumulative effect concerns me most. If everyone uses AI for routine communication, we create a world where most messages are machine-to-machine, with humans as supervisory checkpoints. The texture of human connection through language starts to disappear.

A Framework for Deciding

Context matters. Here's how I think about it:

For purely transactional messages, where the relationship is functional and the content is routine, AI assistance seems fine. Confirming appointments, sending requested information, acknowledging logistics. These messages would be identical whoever wrote them.

For messages where the relationship matters, where your particular voice and attention are part of what's being communicated, AI assistance is problematic. Personal correspondence with friends, messages of condolence or congratulation, any communication where the recipient is entitled to expect your presence in the words.

For professional communication, the answer depends on context. A sales email was probably never expected to be personally composed. A message to a colleague about a challenging project might be. The question is: what is the recipient reasonably expecting to receive?

Tell me more about authenticity and AI

Disclosure as Solution?

One proposed solution is disclosure. Tell people when AI wrote your message. This preserves honesty while allowing efficiency. The problem is that disclosure changes the message. "I used AI to write this" communicates something about how much effort you chose to invest. The recipient now knows they weren't worth your time.

This might be fine for some messages. For others, it's worse than either sending AI-generated text without disclosure or writing it yourself. The disclosure itself becomes the insult.

Maybe this is the point. If you're embarrassed to disclose AI assistance, perhaps that's a signal you shouldn't be using it for that particular message.

The Deeper Question

Behind all this lies a bigger question: what do we value about human communication? If it's just information transfer, AI is an improvement. Messages become clearer, more efficient, faster.

But if communication is also about connection, about the knowledge that another mind engaged with yours, about the small inefficiencies that signal effort and care, then AI assistance isn't neutral. It's a substitution that changes what's being exchanged.

We're entering an era of distributed authorship, where the line between human and machine contribution becomes impossible to draw. This isn't inherently good or bad. But it requires us to think carefully about what we want from communication and what we're willing to give up for efficiency.

---

I don't have a simple answer. I use AI for some writing tasks and not others. The distinction I make is about whether my personal attention is part of what's being offered. Sometimes it is, sometimes it isn't.

What I'm confident about is that the question deserves thought. The default of accepting whatever AI makes possible, without considering what we lose, is how we end up in a world we didn't choose. Better to decide intentionally what role AI should play in how we communicate with each other.