AI is one of the most polarizing topics online today. Scroll through social media, and you’ll find tons of “AI makes me sad” essays: about jobs disappearing, about art being flattened into slop, about something human being lost. I understand where those feelings come from, because I’m tired of reading articles that sound AI written too. I’m not here to argue those feelings away. My views, as a blind person, are very different. This world, both physical and digital, was made for non-disabled people. Computers come with monitors, not Braille displays. Screen size and complexity are put way before speaker size and quality. And accessibility is thought of last, if at all. AI changes all of this in the digital world, and the physical world is next. In this post, I’ll talk about how it’s changed the lives of so many blind people, for the better.
When I was growing up, the only games I could play with any kind of skill were fighting games. That wasn’t because other genres of games didn’t exist, but because those were the only kind of games that had enough sound design and a predictable set of events, and a simple enough system that I could memorize the menus, character select screens, and moves. I couldn’t read the move list even if it was onscreen, so sometimes I found one just by experimenting.
All this took lots of trial and error. I had to try one menu option, see what it did, then another, see what that did, etc. I had to figure out how to get to what I called “two player” mode and then go through the character select screen with one controller, so that the game spoke the character name without moving on to a fight, in order to map out the characters in the game. So, even memorizing things took a lot of effort.
Then I got a laptop and the Internet. I could then watch gameplay videos on YouTube. This was in about 2010 or so, before ads were everywhere. I was still a teenager, and still figuring out screen readers and the Internet at the time. BTW my first emulated video game was Mortal Kombat 3 on the Sega Genesis, using Kega Fusion. I was so excited that I could play that on a laptop.
Then, as the years went by, someone developed Pokemon Crystal Access. It ran an older emulator that had Lua scripting built in, and allowed us blind people to, for the first time, accessibly play a Pokemon game. We didn’t have to figure out the layout of an area by bumping against walls for every step, or read GameFAQ guides to understand the dialog and game mechanics. We could simply play a game.
In 2023 or so, I was talking to ChatGPT about the future of video game accessibility. Of course, this was 3.5 so it’s absolutely dumb by today’s standard. I thought we’d go from a model that understands text, to one that understands images, hook that up to an emulator, and get accessible games. But that’s not what happened. No, what happened was even better!
In late 2026, we now have a Persona 4 Golden mod, a Minecraft mod, a Disco Elysium mod, a Skyrim mod, a Final Fantasy 1 Pixel Remaster mod all the way up to Final Fantasy 6 (with 12 being worked on), a mod for Mario Kart Wii, and even a mod for Zelda 1 for NES. All this happened from May to October of 2026, plus many more mods I’ve not listed. All this is because of blind people taking things into our own hands, and using AI to open doors which non-disabled developers had, without thinking, firmly shut in our faces.
Large corporations are starting to do a little of this. Google was first to build generative AI-powered image descriptions into their screen reader. Apple laid the groundwork with machine learning descriptions, but now they use AI, finally. But both of these only describe a single image, and we cannot add a second image, or video, to that context. So, the only games that are made actually accessible with that are visual novels. For other games, we still need actual accessibility mods, and only blind people will do most of this. Nintendo has only just added a screen reader to the Switch 2. Making their games playable by blind people, let alone older games, won’t happen for years for Nintendo.
With AI, we’re able to add accessibility to nearly any game we want. All it takes now is building the accessibility model we want for a game like Mario Kart or Zelda, and having AI implement it. I never thought I’d ever be able to play Final Fantasy VI, until I did.
Final Fantasy VI is the first Final Fantasy game I played, besides Dissidia Final Fantasy. I arrowed around the title screen, hearing “new game” and “load game” and “options.” I started a new game. I went through the first boss-like battle with The Whelk, a giant shelled monster, and followed the tutorial text about when not to attack (after dying to it once). I got to the point where there are like three parties you have to control, who are at different parts of the world, and my brain couldn’t handle that. But the feeling of joy that I, a blind person whose only experiences with Final Fantasy were a spinoff fighting game and fan fiction novelizations, had actually gotten through a few parts of Final Fantasy VI, stuck with me.
That’s just for gaming, though. For blind computing as a whole, we have several speech synthesizers talking that haven’t talked for a good 20+ years. Stuff like the old Keynote Gold voice (BestSpeech), Vocal Writer singing, a decompilation of BestSpeech, and that same voice running on my iPhone and Android. I grew up with many of the voices we can now run on modern systems. Hearing them speak again feels like a dream.
Again, no company would have done this. Untangling the copyright junk from companies that failed, were bought, and whose products were never seen again, takes a lot of time, and a lot of calls that won’t be answered. Those voices were orphaned. No one sells Keynote Gold or TruVoice anymore. Even Oralux, the last place you could buy IBM ViaVoice for Linux, has stopped selling it. Now, with AI doing the heavy lifting of reverse engineering, a few determined blind folks have brought these voices to modern hardware. And since they’re now source code, they’ll never be closed binaries that need 32-bit bridges and containers to work again. No waiting and waiting, asking and asking.
We’re even building our own tools, down to the screen readers themselves. It started with the NVDA screen reader, built long before AI, for Windows, because the alternative was expensive. Now, we’re making our own Android and MacOS screen readers, because even though the built-in ones are free, they aren’t good enough. Backtalk is a fork of Google’s TalkBack that makes it faster and brings back features Google removed years ago. TurnOver, a brand-new Mac screen reader, steps in for VoiceOver in the apps where VoiceOver falls apart, like Mail, Safari, and Terminal. One person has made an Elevenlabs music and sound effects generator, along with so many other things on that page, which allows people to use ElevenLabs for music and sound generation without needing to deal with the website. I’ve even worked on a small utility for Android, Waypoint, which tries to read incoming responses from AI apps like ChatGPT, Claude, and Google’s own Gemini, because they won’t. Google, Apple, and the rest of them aren’t going to do that for us. Anything that is open source or has a stable API can now be made into an accessible interface, by the people that need it, and all we need is AI.
With AI, blind people have even encroached upon the sacred Physical World of the Sighted. ScribeMe is an app that works with a phone camera or the Meta Glasses to continuously describe the physical world around someone. On iOS, VoiceOver now has scene description through Live Recognition, which uses Apple Intelligence to continuously describe what the camera sees. Gemini Live’s new Guided Vision is more conversational, coaching you to aim the camera and describing things more continuously than regular Gemini Live. All three of these allow you to optionally ask about the current scene to get more details, like where I left my AirPods, or what box of food I pulled out of the freezer. Contrary to many people’s beliefs, blind people don’t have guardian sighted people assigned to them for every waking moment of the day. Shockingly, some blind people live alone in their own home.
These tools do not replace a cane, guide dog, or one’s own common sense. Two of them are also made by big corporations. One of them was built by a blind founder. I still like Apple’s better, since Gemini can’t even turn on your flashlight, so a blind person has to go to Quick Settings, hearing both Gemini and the screen reader talk over each other. Some blind people live with the lights off, since they don’t need light.
When I was growing up, I could only play about 10 games across Sega Genesis, Super Nintendo, Nintendo 64, and PlayStation 2. Now the number of games I can play grows by about a game a week. Now I sometimes can’t even decide what I want to play next. I’ve gone from reading Final Fantasy VI novelizations from a Fan fiction site to actually playing some of it on my computer. We have six Final Fantasy games we can play and understand, without having to memorize a thing. One day, I want to bring that to mobile devices through emulation. We can now get descriptions of the world around us, and the pictures sighted people send us. Even now, AI has given me more than I could have asked for in my wildest dreams. And it’s just getting started. AI may be changing art, flooding the internet with sloppy code, and reshaping society. For the blind, however, it has opened up the digital world in one year more than sighted developers have in decades, and it’s starting on the physical world, too.