When Writing It Yourself Becomes the Opt-Out
Having AI write for us is not only an option in most apps, but in some it’s now the default.
For instance, in the Superhuman email app, an email that’s in your inbox will be analyzed by the AI, and it will automatically provide several replies that you can choose from with a mouse click.
The apps have reversed the friction. AI writing is now what happens almost automatically. Writing it yourself has become the opt-out, the extra step. As writing teacher Ann Janzer writes,
“It’s tempting to default to AI when we don’t want that mental lift of writing yet another email. Applications practically forcing AI-generated text on us makes the temptation stronger.”
When AI is the default, it’s all too tempting to press the “easy button” and let AI do the writing without ever thinking about what is appropriate for AI and what isn’t.
Draw Your Line Before You Need It
One evening, I was responding to customer service requests. I was tired, and Superhuman offered a perfectly good AI draft in response to a customer service request. I used it and sent it, and then immediately wondered if I’d done the right thing.
We need to have intention in using AI. We need to decide in advance what’s appropriate and what’s not. We need to draw the line on using AI before our mouse hovers over an AI-generated reply.
I’ve drawn that line by asking: Is this communication transactional, or is it one where a human response is expected and appropriate?
Different people can legitimately draw the line in different places. The point is that the line is thought through in advance, not defaulted to as the easy option.
AI Doesn’t Write. It Generates Syntax.
Before we discuss when AI should and shouldn’t be used, we need to understand what AI “writing” is. AI writing is not the same as human writing. It’s not writing as humans do it. It’s not a simple question of replacing human writing with AI writing.
AI doesn’t write in the same sense that humans write. Instead, it strings together predictive words. There is no thinking. What it produces may look like writing, but it’s only syntax strung together.
As John Warner wrote in his book More Than Words: How to Think About Writing in the Age of AI,
“Large language models do not write; they generate syntax.”
With AI, there’s no wrestling with an idea, no emotions, no value shaping the words. Writing is thinking, and that struggle is precisely what’s missing in machine-generated text.
Warner explains:
“What ChatGPT and other large language models are doing is not writing and shouldn’t be considered as such. Writing is thinking. Writing involves both the expression and exploration of an idea, meaning that even as we’re trying to capture the idea on the page, the idea may change based on our attempts to capture it. Removing thinking from writing renders an act not writing.”
The default of AI writing isn’t free. It quietly removes the thinking.
AI expert Ethan Mollick wrote in his book Co-Intelligence,
“When we use AI to generate our first drafts, we don’t have to think as hard or as deeply about what we write.”
Where I Happily Let AI Do the Work
It’s acceptable to use AI for transactional communication. In fact, anything that AI produces is itself transactional because it’s not human writing. There’s no mind behind it, no thinking behind it, and no human behind it.
Transactional is the automated stuff: responses to business emails, customer support, and the machinery of getting things done. I have no qualms about using AI there.
For example, I don’t have an issue with using AI to respond to support emails. For instance, a customer writes to tell me they can’t access a Field Guide in the MacSparky Labs Circle Community. “I added that. It should work now. Let me know if it doesn’t.” It is merely transactional, reporting what was done or what’s needed.
I feel the same about using AI to help with editing after I’ve written the first draft of a blog post and made some editing changes myself.
I use AI not to generate my first draft, but to improve what I’ve already written: headline options, section headings, better grammar, and suggestions for closing sections.
What matters is that I don’t blindly accept anything AI suggests. I evaluate it. Sometimes I rewrite it, sometimes I combine it with my own text, and sometimes I just reject it. The judgment stays mine.
So much for the transactional side. Human writing is a different matter entirely.
The Writing You Should Never Outsource
Human writing is for moments where a human response is not just expected but owed: a eulogy at a funeral, words of support for a friend going through a tough time, a thank-you note; anything where personal context is the whole point. You don’t outsource those.
It’s for creative work, like a blog post or a book. My blog writing is human creativity. It’s an expression of my thoughts and feelings, a reflection of my thinking and struggling with a topic. It’s my desire to connect with other humans and help them.
This is off-limits to AI. I therefore don’t use AI to write first drafts for blog posts because I consider that a human activity.
If you want to go deeper on why the human side is worth protecting, read my blog post AI Can Mimic Writing, but It Can’t Replace Human Writing.
When AI Writing Stops Feeling Alive
I respond differently to AI use depending on whether I recognize something as transactional or human.
If it’s a transactional message that I suspect was AI-written, I really don’t care, and I understand why someone would use it.
On the other hand, if it’s a message I’d expect to be personal (from a friend, support in a hard moment), and then I discover it’s AI-written, it really rubs me the wrong way and feels much less valuable.
I don’t want transactional communication in those situations. I want to hear from another human being.
As Oliver Burkeman wrote in his newsletter, The Imperfectionist, Navigating by Aliveness,
“When I realize something’s AI… it no longer feels so alive to me. And that this change in my feelings about it isn’t irrelevant: that it means something.”
Decide Before the App Decides for You
The apps aren’t going to stop making AI the easy path. That part is out of your hands. What’s in your hands is the decision you make before you ever hit reply: is this transactional, or does this deserve a human communication?
Draw that line now, while you’re not tired and not staring at a pre-written draft.
Mine is transactional versus human, but yours can sit somewhere else — what matters is that you chose it on purpose.
The next time an app hands you a ready-made reply, you’ll already know whether to use it or write your own.
Note: I wrote this blog post myself using my own words and thoughts for the initial draft. I used AI only to suggest headlines, section headings, images and text improvements.
Links to product pages on Amazon include a referral code which pays me a small percentage of the sale when products are purchased. This helps to defray some of the costs of running this site. I strive to only include links to products I believe are worth purchasing.