A human was here

On writing humanly

A lot has been said about the limitations of using AI as a tool for writing. In the Atlantic, Jasmine Sun explains how AI largely fails at being creative or interesting. In his blog post titled “Infinite midwit”, Adam Mastroianni draws a dichotomy between objective and subjective intelligence; the latter of which AI lacks. Another Substack blogger, Erik Hoel, has argued that while AI has been great at helping us write more, it has largely failed to help us write better (for example, since 2023, the number of books on Amazon has exploded, but the average review score has declined). These posts are all absolutely worth reading (I try not to share links to things not worth reading).

However, I have seen relatively little attention given to how AI has changed the writing of people who aren’t using AI for their writing. This is surprising because it has made such a difference to me personally. Take a few examples:

1. My prose has gotten significantly more loose. (AI might prefer that I adhere to the principle of concision and instead use the word “looser.” Thanks, but no thanks, AI. My blog, my rules!)

2. I remind the reader that I am writing from my perspective. For example, I say “I think” a lot more now. This breaks another rule of writing: don’t make things about you when you don’t need to. Although I agree with this principle overall—especially when you are writing an argument essay in which it is a given that you are writing from your perspective—I don’t mind reminding the reader that they are reading from a person with experiences and stories.

3. I am allowing myself to write more…informally. A few years ago, I took some pride in my attempt to craft “the perfect sentence.” Now that a 7-year-old can do the same thing by using ChatGPT, I just say what I mean.

The best way to signal competence and trustworthiness used to be to sound authoritative, objective, and clear. The goalposts have shifted a bit since AI has taken over much of humanity’s writing. People have become more skeptical, looking for cues of AI writing, such-as-but-not-limited-to:

1. Adverbing the reader to hell.

2. Following all rules all the time, like a good student trying to please his or her professor.

3. Meaningless metaphors—“It’s not just this, it’s that!!”

4. Does not properly “read the room”, whether that be academics seeking thoughtfulness or blog readers expecting more personality.

5. Mastroianni put it well in his post: “...And so I check, from time to time, whether the leading AI models can do me better than I can. The result sounds like a version of me that has sustained blunt force trauma to the back of the head and spent years recovering in a hospital where the Wi-Fi, for whatever reason, only lets you log onto LinkedIn.”

Credibility doesn’t just come from signals of confidence anymore. It now also comes from giving people the sense that you wrote what you wrote, and that you are a sentient being trying to make sense of the world around you. One response people have to this newfound skepticism is to anxiously delete every m-dash from their writing. Another is the effect it has had on me: I write more honestly now. Instead of trying to take myself very very seriously, I allow myself a few hedges here and there. I allow a personal anecdote. If I am not certain about something, I just say so. Although these are all violations of writing persuasively, it signals something that most of us haven’t had to signal before: a human was here, thinking, writing this thing, having memories, trying to articulate an argument which may or may not persuade you (and that is okay). What matters now is that you have good taste and some charisma. AI will not help you with that.

Does this mean we should jettison all of our previous rules for good writing? No! Many of those will stand the test of time, such as:

1. Don’t just say; show. (tell a story!)

2. Use active voice

3. Avoid cliches like the plague (this rule has ironically become a violation of itself)

4. Omit unnecessary words (this rule fittingly epitomizes itself)

5. Remind yourself that you are writing for someone who is not you, and might not know the things you know

6. Short is sweet

7. Be honest

Honesty is perhaps the rule that guides good writing the most. It forces the writer to say what he means rather than saying what he thinks he should say, or what others might say if they were writing what he is writing. I wouldn’t recommend someone indulge themselves in a 5-page rant, but reminding the reader that there is someone behind the text—who has stakes and a reputation to manage—is perhaps a stronger credibility signal than anything else.

For me, writing has become less about the art of being useful to the reader, and more about… well, that, but also allowing some subjectivity to creep in while I do it. While AI loyally abides by every rule of language, humans creatively break rules, improve on things, and inspire future writing because of it. Now that everyone can write in the prose of C.S. Lewis, there has to be some other metric by which we can distinguish good writing (thinking) from bad writing (thinking). That is by having interesting thoughts, connecting them together in a sensible way, and showing off your humanity while doing so.