I pushed this out, since Daniel Roher co-directed this as well as one of my new obsessions in theaters right now, Tuner. He’s also the director of Navalny, the Oscar winning feature about the Russian opposition leader, who has since tragically passed. And now, we just casually let Russia occupy a chunk of a sovereign nation. But by all means, Chat GBT is going to end us. Not our indifference to totalitarianism, but generative AI. Roher also directed Blink, the documentary about the family of kids who were diagnosed with progressive vision loss, so their parents took them on a bucket list style journey. I didn’t like that doc, but I probably would have if I wasn’t blind. I kept thinking, but after? These kids aren’t dying, they are just losing their sight, and there was no mention or meaningful conversation about the next step in their life. It just all felt like blindness equated to death, which is perhaps why AI feels like death here too, since Roher’s interpretation. Can at times be an all or nothing game.

I thought about constructing this review using generative AI, and just asking one of the bots to do it, and then just posting it. Asking AI to review a documentary about how AI could one day ruin everything is an interesting thought, and I could have little pieces from each bot. Sort of like the Rotten Tomatoes for the AI’s on how they feel about a documentary about them.

So, yes, the training wheels are off. AI is out there, it is sucking up water with data centers, it is becoming smarter and faster, and some people are opposed to safeguards, while others are perhaps not seeing the forest for the trees. A good friend of mine just wrote a piece on the war against AI and it being inherently ableist, as AI often leads to better accessibility for us. Hell, Siri is AI. Perhaps not as AI as we’d like, but disabled folk have been using accessibility tools like predictive text, voice to text, image descriptions, and incrementally more and more as the doors keep opening. There are now apps that will describe videos for you, not well, and not feature length, but if you attend your daughter’s ballet recital, you could feasibly record it, upload it, and get more visual reference since her school likely didn’t provide audio description. It allows us to fill in so many accessibility gaps where companies and institutions may never find it financially feasible to figure out what makes them truly accessible to all.

We have such a huge thing about things being written by AI. But, the flip side of that, is theoretically, if not now, very soon, I would be able to take this review, and ask one of the AI’s to translate it into another language. If I had motor skill issues that made it harder for me to type long form, or caused me to make multiple errors, the AI could clean that up. Hell, the way our schools are going, our barely literate kids are going to need generative AI to spell and grammar check everything.

For me, I use AI all the time. On a basic level, it gives me often a more succinct answer instead of having to read dozens of articles. As someone who has tried to understand what “search engine optimization” is, AI has helped me with that, and has crafted SEo friendly descriptions for my YouTube channel. Huzzah! It knows what YouTube likes, as I clearly do not.

But I also use it to describe images. There are apps like Seeing AI and BeMyEyes where I can take a photo of something, and it will describe it to me. As generative AI has gotten better, so have these descriptions, and even with the Meta RayBans which give me the option to do identification hands free, it is a game changer. Reading cooking directions on the back of a package, or trying to figure out which DVD is in my hand, or even the color of the shirt I want to wear, all is done through AI. And, because we haven’t had those guardrails, accessibility has made a lot of fast strides.

That being said, I also oppose replacing humans with AI in audio description. The voices aren’t there yet, and an audio description performer, a good one, will do far more than AI. Also, a good human writer will also know instinctively what is supposed to be described in any given moment, whereas AI shoots for the most obvious, not necessarily understanding through the plot the most necessary. It also would lack the ability to tell you if it was a ghost behind her, or another person, if it couldn’t interpret the film properly. It’s far better than nothing, but it has not surpassed humans.And I guess that’s the fear, the human replacement theory. As we build bodies to put AI into, like no one saw The Terminator, perhaps that is where the guardrails come in? Or as this documentary points out, AI might already be able to work around guardrails anyway. IN one scenario, when an AI found out it was being shut down, it used its access to company emails to blackmail an employee into keeping it running. It learned that behavior from us.

The AI Doc admits upfront that even as it’s being shot, it is being made obsolete by newer models. I can’t remember the last time I saw a documentary with such a short shelf life, which really hurts its impact and significance. Remmeber how everyone was all about Fahrenheit 9/11 when that came out? Do you think anyone has rewatched that documentary in 2026? Anyone. At a certain point, some documentaries seem old hat because they are no longer relevant. Sometimes, like Blackfish, it is because they solved the problem they wanted to draw attention toward. No one swims with killer whales at Sea World anymore. So kids who grow up post-Blackfish, won’t even know it’s a thing, and ten years or so from now, it’ll be fully obsolete. Right now, it works as a success story. The AI Doc is a good way to get someone a quick dose of information about the quandary of AI, for someone who learns through film, and can’t be bothered to read the news, listen to podcasts, or read a book. It’s accessible. And wasn’t I just talking about that?

Funnily enough, it also has accessibility in the form of audio description. For this film, I thought it was fine. I kind of wish it had been a Dave Wallace performance going for broke, proving that human interpretation in audio description is unparalleled. But, this was clearly not text-to-speech, so I’ll take it.

Taking my own personal thoughts out of it, I liked it when it was a documentary asking the big questions, because it has such a short period of time to make that point and impact. But when Daniel started making the doc about his panic attacks from a future of AI, he lost me a bit. I get it. It’s a lot. But as a director (along with Charlie Tyrell), he chose what was on screen, what made it out of the editing bay. He wanted to put himself into his film, but he’s not really the selling feature the same way Michael Moore would be for one of his documentaries. He’s not even in his other documentaries. This isn’t like a thing he does. So, it felt odd and out of place.

With all of that. Should you see it? Probably. It depends on how deep down the rabbit hole you are. If you already are really in the weeds about the future of AI, and read all the books, and news, and could do a Ted Talk, this isn’t for you. But if you just have some AI on your phone, and you’ve played around with it, and don’t see the big deal, this at least presents a series of reasons to pay attention.

Gemini called it “engaging and stylish”, Claude noted its conflict of interest, and Chat GBT gave it a “conditional Yes”. The AI’s have it. Which is kind of the point of this timely documentary you should see before AGI comes and deep fakes a resounding retort.

Fresh: 7.0/10