A few weeks ago I wrote an article about the Photoshop clones (“Artcraft”) that were almost entirely coded by AI. I was surprised that this was already possible, and impressed by how well the software worked and how quickly it progressed.
It made me install a free AI coding app too and see for myself.
And what did I see? That it would be stupid to ever manually code something again.
The Experiment
For years I basically ignored AI developments because I was faster and better myself. Oh, AI can autocomplete my code? No thank you! I can type just as fast and be sure it’s 100% correct.
But now you can just tell AI to implement a feature. It will scour your codebase, understand it, and do it for you. I kept testing bigger and bigger features, more and more complex tasks, and it just … did what I asked 90+% of the time. The only times it screwed up was when I used a much cheaper year-old model (because I was on a free account with limits!) or when my instructions (or project specifications) just weren’t clear enough. So it was my own fault half the time too.
I tried it on games. I tried it on websites. I tried it on software. And it could do all of it, albeit slowly and restriced because I used free accounts.
At first, this gave me a bit of extra energy and motivation. I thought about all those great ideas of mine that I could never make because I lacked resources (time, money, etcetera). I could make them now! I could make my “dream game” in a year instead of 10 years. No, wait, I should use it to create paid web services, that might actually earn me some solid income.
I brainstormed apps and websites I could let AI create. I created lists of things that used to be out of reach but were very much within reach now. I did more research into AI and market research into paid web services …
… and I realized—which was quite obvious in hindsight—that everyone else could do the same. Anyone else can prompt AI to create some web service. Even worse, ayone else can get AI to create a little tool that does what they want. No need to ever pay for anyone else’s tools anymore.
But I try to be a positive guy. I try to find solutions instead of get stuck whining about problems. So I pivoted to educational tools.
I’ve been trying to get a large online educational store going for ~2 years now. Creating educational books, puzzles, escape rooms, games, lesson plans, worksheets, etcetera. With AI, I could create really strong tools for creating these. Not just for myself, but to give to teachers / students / homeschooling parents too. A tool to quickly create tests. A tool to rapidly grade something.
But then, you guessed it, I realized everyone else has already done the same thing. Ever since AI became good enough for real coding—which is about 6+ months ago—the number of online services to help teachers has skyrocketed. These tools are mostly just a thin layer for AI. A teacher basically pays triple for what they could also do themselves, directly, while staying in full control. But many people don’t know that and the first services to offer this are very big and profitable now.
I would like to say “these tools are terrible” … but they’re not. I’d like to say these services are scamming teachers, or that I could make something better myself, or that AI in the classroom is some horrible abominiation … but it’s not. Most of these tools are literally just a layer between you and AI, but some of them have grown into actually useful toolsets that simply incorporate AI at many points.
Recent studies and surveys show that more than 2/3 of teachers (in America and UK) are using AI on a daily basis for their work. And I can see why now. With a single prompt, you get a pretty worksheet or a nice graphic explaining your topic. With one or two prompts, you get a perfect lesson plan, or you get 15 lesson plans perfectly tailored to each individual student of yours.
I can’t compete with that, at least not with manual work.
Can I compete with it using AI?
Adding AI As A Tool
That’s what I’ve been researching the past few weeks. I’m a very experienced programmer and framework builder, and an experienced teacher and tutor. Surely if I use AI coding myself, I can create something that’s way better and that teachers will pay for?
Well, my research reached quite a clear conclusion: no, I can’t. Here’s why.
- Most free AI tools are good enough for most of what teachers need. ChatGPT’s images have been good enough for a while now to just get what you want first try, for free, easy-peasy. There’s nothing to do for me there.
- But free AI tools are not good enough to create serious projects. I really tried. I have like 5 different accounts now and burned through all their free usage limits (for this _month) within a day or two. Despite writing very clear specifications for products, despite my most clever efforts to reduce usage, I just could not get to the finish line this way.
- ChatGPT offered me one month of “Premium” for free. I took it, eager to check if it made a difference.
- It did, but in all the wrong ways :/
- The $20 subscription barely gives you any higher limits or better models. It’s still far from enough.
- At the same time, it does give you near unlimited image generation and chatting. Since learning this I have also learned that tons of teachers and parents have this subscription for exactly that reason.
- Consumers by and large don’t care about AI. Many can’t even spot it. Most don’t care at all. They don’t have ethical concerns and they have no taste.
- With that I mean to say: standards have been declining for a while now and what used to be “not good enough, be better!” has become “meh we’ll accept anything that kind of works”.
- I briefly considered pivoting my store to “the only educational store without AI” or “human-made resources only” or something, but saw no evidence at all for why it would help sales.
- Whereas I, as a creative, feel hollow and unsatisfied by using AI. After an entire day of prompting, steering, correcting, fixing a handful of code bugs myself … I felt restless. I felt nothing. I had not actually done anything. Though I had achieved massive results, they meant nothing to me. (And I had to live with the uncertainty of whether it actually worked or what bugs might be hiding everywhere, because the codebase wasn’t 100% mine anymore.)
Why manually create worksheets when teachers will just prompt AI five seconds before class starts? I can’t blame them. Free, quick, good enough. Maybe even better because you can prompt it for a worksheet that matches your class exactly, instead of searching the internet to find a premade one that only matches somewhat.
Why create educational quizzes and stories and puzzles? When a teacher can let AI read their textbook and turn it into a quiz instantly? Free, quick, good enough. And it probably fits even better than the more general/broad quiz that I would make.
I was under the impression that AI was used by some teachers. And that they were just using the free version for the small benefits it offered. But now I’ve done research, asked around, talked to some more people, and my eyes were opened to the truth. Entire schools are mostly run by AI now. Teachers actually have paid subscriptions to get all the extra benefits, sometimes paid for by the school.
So I’m not just competing with someone using ChatGPT for free. I’m competing with teachers and parents who are almost subsidized to use high-quality AI plans that give them near unlimited tools.
I don’t have the money to spend $100, $200, $500 a month on the best possible AI subscriptions. The ones where I get enough usage to create tool after finished tool and really compete with what’s out there. The ones where I can prototype, iterate, and release fast enough to stay ahead of this curve and maybe become the best in some department for a few months. Or the one that will allow me to create one of my many “dream game ideas” that I’m pretty sure lots of people will want to buy and play.
I could create my own layer between user and AI. For example, you ask users 5 cents per “tool usage”, and I pay 2 cents for that in API token costs at the backend. This technically doesn’t need upfront payment. (I only pay a fraction of what the user has already paid.) Which is exactly why it’s the dominant business model and we’re already drowning in services like these, both for the general public and for teachers/schools.
And those other business are way more ruthless in what they’re willing to do than me. You know, a person with concerns about climate, ethics, educational standards, and handing our thinking to AIs.
So now we’re in a situation where …
- Most people have access to the same tools as me. Even worse, they can access better ones. (They are subsidized by schools or simply use their income to get higher-tier subscriptions. This isn’t limited to some small pool of teach-savy power users—it’s everywhere already.)
- Everything I worked hard to become good at can now be done “good enough” in two seconds by anyone. (Most AI images teachers generate and give to kids make me want to gouge my eyes out, but they don’t care and the kids don’t care because it’s “good enough”.)
- I can’t really use this AI to innovate and create something uniquely sellable. (I have superior skill and effort, as making creative products has been my full-time job for a while now, but I lack the resources.)
So What Now?
I noticed something unusual after my first week using AI.
The week had pretty much flown by. I had barely gotten any other work done. I’d been prompting AI, checking its result, writing clear specifications on what to do next, then letting it do the work. A cycle, over and over, that helped me create a massive tool for teachers. Until, with ~30% of the work done, my usage (on a premium account) was gone for the week.
It forced me to spend a few days away from the AI again. To finish my other projects manually and focus on some things I’d ignored all week.
And there was … nothing. Normally, by the time the weekend rolls around, I have a massive list of little ideas and tweaks and creative thoughts. It’s a habit of mine to use every Sunday to clean up this mess and get that pile of creativity sorted somewhere.
But this week ended with zero new ideas. Not even a little melody in my head, or a fun title for a book, nothing.
The next week, AI helped me get that massive tool to ~60% done, then usage ran out around Friday. (I was learning and becoming smarter about the usage and prompting!) On Sunday, again, I had … nothing. No ideas. No notes. Nothing written down on the sketch paper next to my laptop.
I was both tired and restless. I had achieved so much, yet so little. And using AI this intensely seems to completely shut down my brain. The creativity is gone, but also just the general willingness to think and to work through problems. It’s scary how quickly you are trained to think “I’ll just prompt AI to do it tomorrow” instead of “let’s do that myself now, it’s 5 minutes of work”.
I can usually get a lot done in a week. I had a nice plan to finish a few projects, step by step, and I knew it was absolutely doable. But using AI simply made it so hard for me to do any of that work, that I only got one part of that to-do list done. In a week. Taking a tiny step each day when I had finally convinced my brain to activate (and had to wait on the AI to finish anyway).
So what now?
I can continue to barely use AI. Only as a tool in very specific situations where it makes sense. It would mean falling behind everyone and everything, while never producing something as tailored and quick as what everyone can already get for free.
I can start using AI intensively. Put all my budget into a great subscription and have at it. It would be a questionable financial decision and it would shut off my brain and creativity.
I can continue my educational online store. Trying to get a foothold, show how much better the system could be, design something that will really help make the lives of teachers and students easier. Focus on that “small” sector (instead of the general public) and do whatever it takes to get ahead of AI and existing paid tools. Just accept AI does 95% of the actual work now and try to work around that. For example, just let AI churn out worksheet after worksheet all day, put it online, pray.
I can drop all of it and pivot to something else entirely. The reality (habits, market, competitors, teacher preferences) is stacked against me. I should switch to something where AI is less powerful or way less accepted/used. Focus heavily on live performances, such as music or book readings. Focus heavily on physical objects like board games or hands-on work.
I … don’t know.
The past few weeks I’ve learned so much about AI and all the people using it. I’ve played with it, experimented it, interrogated it in all possible ways.
For years I was able to say that I could do something better, or faster, or more creatively. I was able to say that I’m always growing and creating better work, and that this will eventually lead to a stable income.
But now I can’t say it anymore. Our AI tools are good enough to do almost anything involving digital or media work. They can do the objective stuff (menial work, administration, grading) as well as the subjective stuff (creativity, storytelling, images, games) in equal measure.
An AI coded game has no soul and making one would make me feel hollow. But half the people don’t see it at all, and the other half buy and play the game because it’s pretty good.
A curriculum of AI generated unchecked nonsense is terrible for teaching and a terrible example for kids growing up. But 2/3 of teachers are literally doing this themselves, and the other 1/3 are likely paying random webshops to buy their massive bundle of AI-generated worksheets for $5.
It’s the old story of ideals clashing with reality, of capitalism clashing with quality, of the artist fighting for a way to earn an income, of innovation leading to loss of jobs.
It was hard before the advent of AI. It’s downright impossible now that it’s here. The only way to keep up is to become an AI power user myself … but I can’t become a power user with my limited budget. (And strong inner desire for creativity and making stuff with my own two hands.)
Conclusion
So what now?
I don’t know.
I’ve been fighting with this for weeks now. Months, really. I’ve been aware of AI and its progression for much longer, of course. And I keep reaching the same somewhat nihilistic conclusions.
All I know is that things are changing and things in my own situation need to change with it. I set my online store back to “test mode” (completely usable except buying) because I just wasn’t selling any of it, while paying a hefty sum (for me) every month. I already subtly let more and more AI creep into my work, thinking I need to be faster or that it would somehow reveal insight into where I should go. I already moved a lot of my work to other platforms (where teachers/schools buy things), while stepping away from the educational angle and working on some other things that AI can’t reproduce right now.
I’ll probably keep moving in that direction until I find some answers. Maybe I’ll end up back where I was but with renewed insight and motivation to use AI in the best possible way. Or maybe I’ll end up all the way on the other side, become a carpenter or something, never do digital art/products again. I increasingly feel like all this creativity should just become a hobby, any idea of making income or making it my job banished, and hope to earn income elsewhere—and hope that actual fulfilment from making stuff returns.
I don’t know.
But this has been on my mind, this has been what I’ve been doing, and I just wanted to share those thoughts.
Tiamo
