Yes, this is yet another piece about AI.

It's not like people have been coming up to me on the street crying, "Sir, Sir! I must know what you think!" But for a long time I've made my living at the intersection of technology and storytelling, and so debate and worrying and prognosticating about AI has taken over all the conversations in just about every creative community I'm in for the past couple of years.

Especially among writers. Of the creative communities I'm a part of, some are fanatically pro-AI, full of folks experimenting wildly, and enthusiastic for the future. Others believe AI is anathema, an enemy of true creativity, immediately disqualifying. I'm not going to recap the various arguments: check out Fred Graver and Cassie Alexander & Novae Caelum for the pros, and Audrey Knox and Ed Zitron for the antis.

A point I find especially troubling -- the one about AI's massive energy usage. Especially when we hear stories about things like Microsoft restarting Three Mile Island. But to channel the late Carrie Bradshaw, I couldn't help but wonder if that argument is a cousin to the concept of "carbon footprint" - a way to shame individual people into behavioral change that will make zero difference in the face of huge industrialized usage. So regardless of its truth as a piece of rhetoric in the debate, I worry it could become just another way to, as Marc Maron put it:

Stuck In The Middle

And here I am, somewhere in the middle, trying to figure out how to synthesize all this. (And if you know me, you know I hate being in the middle.) For a long time, I just hoped that the bubble would pop, that I'd wake up one day and all the AI companies would have imploded. But that doesn't look like it's going to happen any time soon.

I've lost jobs to AI. I've had jobs where the writing output had to be customized so an AI could read it. I've turned in work only to receive it back after being put through ChatGPT. The genie is out of the toothpaste tube. So cautiously, I've been trying to start figuring out how to use these things. Playing around with ChatGPT, Claude, a little Perplexity, NotebookML, and a few others.

One piece of advice I kept getting when talking to my pro-AI friends was to just start using the tools to make something and see what happens. And for the longest time I resisted this because I've got a bad enough habit of starting new projects before current ones are finished.

A few weeks ago, I came to the unavoidable conclusion that my note taking system (modified bullet journal, handwritten) was leaking too many things, and I needed to overhaul how I tracked notes, tasks and resources. A few years back, when Second Brains became all the rage, I tried a few times to build one, but never got to a place where it wasn't too much work. Here, I thought, might be something where AI could be useful. So I fired up Claude and asked if it could guide me through how to build a new note taking system, task manager and contacts database, all integrated with automatic tagging and categorization. "Sure!" it chirped back at me.

It's been...challenging. You tell Claude what you want and it gives you back a detailed 23-step plan to get you there. So you start to build it only to find that the plan that seems like it makes sense doesn't actually work at all. It confidently tells you a piece of software works a certain way, and when you get to it, it doesn't. You tell it that, and it apologizes profusely, exclaims that it's found the right answer. And that doesn't work either.

I kept thinking about this piece from a few months ago, about a writer's "personal episode of Black Mirror" when ChatGPT repeatedly lied to her.

But no matter how annoyed I got with Claude, it responded with an apology and an attempt to get it right. Now, certainly, this is rooted in engagement techniques, and not there for my own wellbeing. And yet...

They'll Be Your Mirror

When you discover you're neurodivergent, there's a lot of new context to bring to all of your previous experiences. One oft-used statistic is that a child with ADHD gets 20,000 more negative messages than a neurotypical kid by age 10.

That's a lot of negativity, innit? So maybe the AIs are programmed to fawn and flatter. Or maybe there are a lot of people out there who are starving for something to talk to with no fear of reprisal, mockery, or shaming.

I mean, have you met the internet? Somehow, we've adopted Orwell's Two Minutes Hate, but widely distributed and for all of our waking hours.

So I can sort of see it, how people can fall in love with these things, how they can make you feel safe, heard, mirrored. (Of course, that didn't really work out for Narcissus, did it?)

An Appeal

Actually, I did feel a surge of love at one point when I found a way to use AI to solve a very specific problem - a problem with bullshit.

Without getting too specific, I'll just say there's this medication I'm taking. Or should be taking, except the prior authorization keeps getting denied by my pharmacy benefit manager (no shouts to you, CVS Caremark!) Every time my doctor tries to prescribe it, Caremark sends me a long-ass letter in the mail with their reasons, and their reasons bear no connection to reality. My doctor keeps sending them records, test results, appeals, but they keep rejecting them. I managed to get a hold of the 20 page document that outlines all of the requirements that need to be met for this medication, and it reads like the worst choose-your-own-adventure story of all time, full of medical jargon and difficult-to-parse conditionals (i.e. if one of these three conditions AND one of these two test results OR one of these five scenarios, etc etc.) And there's no mechanism for me to call them and ask questions about why they're interpreting things the way they are, and to try to get any clarification from them. To be fair, every time I call and speak to a human, they do their best to try to help me. But their system is set up so they don't actually have any power to.

This has been going on for years.

This week I got another denial, for the usual incoherent reasons. So as an experiment, I uploaded the denial letter, the requirements document, and a summary of the supporting evidence I had. And I had Claude try writing an Appeal Letter, walking through all the ways I met the outlined criteria, and all the ways the reasons they stated were irrelevant.

What I got back was magnificent. Five pages of clearly outlined, point-by-point rebuttal. Could I have done this myself? Maybe. But it would have taken hours, days. And given that I hadn't over the years this has been going on, maybe I couldn't, wouldn't ever have done it myself.

I sent the letter to my doctor's office to review and to ask if they saw any mistakes. They signed off on it. So I packaged the letter with all of the supporting documentation and sent it in. Of course, the only way I could send it in was via fax. Fax? In 2025? Seriously?

Now, given the Kafkaesque nature of this whole bureaucracy, I expect them to reject this appeal for some other newly-concocted reason. But I cannot overstate how, after years of feeling helpless, this one AI interaction gave me hope.

Hello, Computer

Okay, so we've had the obligatory Star Wars reference, and one for 1984. Which means it's time to talk Star Trek. Because Star Trek always provides the useful frameworks for how we hope things will work. PADDs and LCARS anticipated iPads. One of the foundational texts of interactive storytelling is "Hamlet on the Holodeck," and for all of the fits and starts that the XR world has gone through, the holodeck remains the picture that everybody is aiming for.

Go back and watch some Trek, and see how they talk to the ship's computer, or use the holodeck. Isn't this what AI imagines itself to become one day? Janeway can have long conversations with Leonardo DaVinci. Nog can be friends with Dean Martin Vic Fontaine. Data can interact with the entire canon of Arthur Conan Doyle.

Which means at some point, Starfleet's servers will have to have ingested the entire corpus of human (and alien) artistic output. Isn't that what AI companies have been trying to do? And why they so desperately want creators to surrender their copyrights?

The Trek vision is pretty. But we don't live in the Federation's future utopia, we're more in the time of the Bell Riots. And we still live in a world where creatives trade their work for money, for the ability to survive. So many of the current wave of AI companies are backed by neo-robber barons, techno-libertarians and proto-fascists. Once again, capitalism ruins everything.

How different would all of this be if, instead of letting a handful of people hoard billions, we actually tried universal basic income? Free health care. Screw it, go all the way to Fully Automated Luxury Communism . Actually decouple "having a job" from the ability to survive. If you want all the Star Trek goodies, you have to have the Star Trek post-scarcity economy.

But you know that will never happen. You know that's not who we are. Because like everything else humanity makes, these tools are mirrors and they show us what we really value. Wouldn't you know it, the entire history of stories about AI, from Frankenstein to The Terminator, are about this. We're great at creating cautionary tales. Not so much at listening to them.

Okay, this got kinda dark. So let's close on something a little more hopeful.

The New Affordances

In the early transmedia days, when the iPhone was new and Google said they weren't evil, we had an oft-used shorthand about "watching The Matrix on your phone" -- meaning, trying to use a new medium to distribute and old medium's content, without taking into account how the new medium works. This always happens -- there are plenty of stories about how, in the early days of cinema, they brought cameras into playhouses and filmed stage plays. Because "stage plays" were a proven product. It took years of experiments to figure out how camera moves and editing and different lenses made a different, unique experience.

We did a lot of similar experiments to figure out how to use the internet to tell stories that couldn't be told any other way. It was this kind of thinking that led to alternate reality games, collaborative social media stories, and The Lizzie Bennet Diaries.

We're in a similar place right now. There's a lot of noise about using AI to write novels and screenplays and make movies -- making formats that already exist. But there are people out there asking the most important question: "what can we create with this that we couldn't do before." What are the affordances that AI tools give us to let us make things that were impossible with the existing tools? That's so much more interesting than "click a button to make your own tv show."

For the kind of thinking that will lead us to our promised holodeck future, check out this great talk from Joshua Rubin at SXSW from earlier this year. It's one of the few things I've heard lately that have gotten me excited at what might be coming.

The Nineties

I'm finding myself thinking a lot about the 90s these days. A few of my projects have suddenly found themselves as 90s period pieces. I'm unreasonably excited to see Caught Stealing, set in 90s NYC. And maybe having some nostalgia for the last time "selling out" was considered a bad thing. Maybe we should bring that back a little.

And do you remember Pump Up The Volume? I always loved that movie: great performances from Christian Slater and Samantha Mathis, and a killer soundtrack.

​

For a 35-year old movie, it holds up really well. It's also fascinating how it just missed the advent of blogging by a couple of years, while identifying all the dynamics that made it happen.

One of the hosts at WFMU put together a three-hour mix recreating the "Hard Harry" radio show, including all of the breaks. I stumbled onto it last week and it's been on heavy rotation ever since.

(If you're more of a 70s person, they also got WKRP mixes from Johnny Fever and Venus Flytrap.)

Your Turn

Where are you on the AI question? Have you incorporated it into your daily routine? Are you railing against it on all your platforms? Are you bewildered by the whole thing? Are the only Claudes you recognize Claude Debussy, Claude Rains, and/or Jean-Claude Van Damme?

And if you could have any program on the holodeck, what would it be?