100% HUMAN

An indie filmmaker tries to prove to a machine that she writes her own newsletter.

Hello friends,

One of the first newsletters I ever wrote was about how I was using my fear of sounding like ChatGPT as a motivational tool to level up my writing. Basically: how to make sure everything I write sounds human.

Now, with Substack’s addition of Pangram, readers can run my text through an A.I. detection filter to find out whether it was fully written by A.I., fully written by a human, or some other extremely precise answer like 76% Human, 24% A.I.

Analysis of this newsletter: phew!!!

We probably need tools like this — they sure are useful to institutions and other well-resourced people — but the problem right now is that A.I. detection tests appear to be about as accurate as lie detector tests.

Which is to say, not at all accurate.

So now, the last step of my newsletter writing process is running it through an app to find out if it is being flagged as A.I. Then I have to guess what I did in my writing to make me fail the test, and rewrite it until I pass.

It’s like a CAPTCHA, but for proving you have a soul.

It’s also a dumb test that’s easy to game.

Once a straight-A student, always a straight-A student.

How it works now:

I write the post, and it’s not until I get to the PUBLISH page that I can run it through the test. The test usually comes up 100% Human but sometimes it’s like, 85% Human, 15% A.I., and I freak out. So I have to go back in.

Sometimes, this results in good edits! I will rewrite stuff that I can admit is a little too generic. I can insert more concrete details. I can use more color. All good stuff that makes it sound — not really more HUMAN, but more like GOOD WRITING, or at least more vivid writing.

The thing I want to point out is that this is a polish pass I might not have the hours or resources to do right now; the newsletter does not make enough money to justify more than a certain number of hours per week on it. But I am a straight-A student, damn it, and sure, if doing the HUMAN test is just forcing me to do more editing: fine.

But sometimes, doing this pass doesn’t get me to 100% Human, and there’s more stuff I have to go back in and change. This is when it gets stupid, IMO. Here I need to look for formatting quirks, or turns of phrase, that happen to be associated with A.I.-generated writing these days. Now I’m just undoing perfectly good writing, just because it seems to be similar to what A.I. writing sounds like currently.

Presumably, those systems learn what “human writing” looks like from reading things humans wrote. Now, humans have to change how we write to satisfy those systems? Also, in case you’re curious: I can ask ChatGPT “make this sound more human” and it does a pretty good job of rewriting to pass the test.

Does any of this make sense??? What are we doing?

Friends: I CHOOSE TO LOL.

I don’t think I should be writing toward a machine that devoured like, all of human civilization, in order to come up with its idea of what 100% HUMAN sounds like. A shark does not become human by eating enough people.

I understand institutions — like universities and edited magazines — need something like this test to see if students are cheating or journalists are not writing their pieces. But for me, do I? Does someone reading my newsletter need it? I don’t know.

This is not a complaint about Pangram specifically. The test may get better. Maybe someday it will become perfectly accurate, or we’ll use some other kind of metadata or watermarking, or maybe we’ll just stop caring about this whole issue and reward writing that strikes us as good and interesting — and simply ignore writing we think is lame and sounds like A.I. Because who cares???

Like: if you think

Yeah! Who cares anyway??

I don’t want us all writing toward the machine-administered HUMAN TEST, especially when it’s so unreliable and just makes us do dumb stuff that is not really about being human. As I wrote back in December, good writing will always sound more human than not good writing, but we should all find our own idiosyncratic ways toward that goal.

And anyway, writers have always adjusted their writing to satisfy various systems, right? Marketing logics, house styles, standardized tests, search-engine optimization, MFA programs, genre conventions… many different institutions have always trained writers to write toward particular standards, through both reward and punishment.

The fun new part is that the machine isn’t merely calling your writing unsuccessful, unfashionable or bad. Now it gets to call you A FAILED HUMAN. For what it’s worth, I know teachers and editors need tools like this, but I can say for myself, having played around with this tool: being told by a machine I am not even a human is very insulting, and makes me feel bad about myself, and doesn’t make me love writing more than I did before that.

I probably won’t be continuing to “check” my own writing against Pangram. I have better uses for the final hours before my weekly deadline, other ways to serve the reader instead of passing this dumb test. Again, I always devote as much time to editing my writing to make it as good as possible, within a reasonable timeframe, on a weekly deadline. If I want to be able to spend more time editing and improving each piece, what I need to do is continue to do good work and hope I earn more paid subscriptions to support the time and effort it takes to do it.

ANYWAY,

Pangram says this newsletter is 100% HUMAN.

But I did have to rewrite a couple parts of it to make that happen.

Comings, Goings, & News

Wild Inside Theatrical Going Strong in Week 3!

We’ve booked close to 100 theaters now. More sold-out shows — I’ve seen people being turned away!!?! definitely a new experience for me — 100% on RT — most importantly, I am just over the moon with the way audiences are showing up, loving it, laughing, crying, hugging their friends who went to the theater with them, asking smart questions and posting their (mostly) thoughtful responses online. I’ve also been really impressed with & inspired by how well the community engagement work is going. As many on this platform are constantly pointing out, local partnerships are the secret to getting people into theaters. And I’m thrilled the film is providing a platform for local birding / wildlife / conservation groups to get the word out about their important work.

Anyway: most engagements are both BRIEF and HAPPENING RIGHT NOW, so do not delay getting a ticket if you are interested in seeing the movie! ICYMI, I share much more of Flaco’s Odyssey — and our odyssey of releasing it — on Instagram. Also: buy merch: I promise you’ll get compliments on it.

You Know I Hardly Ever Do This: Please Support This Newsletter

FREE SUBSCRIBERS: I LOVE YOU! If you ever find the ideas in this newsletter helpful for your studies, influential on your art-making, just as idle entertainment or even something to hate-read (no judgment, haters are welcome!) — please consider upgrading to paid.

It’s only $6/month to subscribe, and it really does help keep the lights on. Paid subscribers get access to the full unpaywalled archive, including my more time-intensive, instructive, and revealing writing.

PAID SUBSCRIBERS: I LOVE YOU EVEN MORE! If you’ve ever forwarded this newsletter to a non-subscriber because you knew they’d love it (or hate it), or used the newsletter to help with a classroom discussion, or just really love my films and want to help me keep making them… please consider becoming a Founding Member.

You can be a Founding Member by upping your annual subscription of $66 to any number over that — like, you can pay $67 and get to call yourself a Founding Member. (At $240/year, Founding Members also get a one-on-one consultation with me.)

Maybe I’ll open the chat for Founding Members only, would that be fun?