Hey everyone,
I’m an online educator running a solo business called Tyler’s AI Powered ESL — Education Evolved, and I wanted to share some thoughts as someone who jumped back into ChatGPT Pro after seeing the incredible March demo of the new image generation tool.
What OpenAI has built here is nothing short of incredible — natural-language-driven, visually consistent image generation that could genuinely revolutionize how we create educational content, especially for young learners.
But unfortunately, it’s currently a massively missed opportunity.
The Use Case: Real-Time Visual Storytelling in the Classroom
I work with Chinese students (ages 6–12) in online 1-on-1 ESL classes. I’ve spent the past few months building lesson plans that integrate:
- Real-time AI narration
- Custom comic panels
- Interactive decision points
- And eventually, adaptive open narratives with visuals
I’m now building an API to handle adaptive storytelling in real time — think ChatGPT-powered choose-your-own-adventure stories where the characters, world, and language dynamically shift to match the student’s English level and interests. Visuals are critical to making that work — and DALL·E 3 almost delivers. Almost.
The Problem: Censorship Kills Engagement
The current content moderation system is so overzealous that it completely breaks basic comic-book storytelling logic:
- Superheroes can’t confront villains without triggering the filter
- Justice and consequence can’t be visualized
- Even mild cartoon humor or visual metaphors (like someone in a trash can) get blocked
This makes it nearly impossible to keep boys, in particular, engaged. If we can’t meet their expectations with even Saturday-morning-cartoon level stakes, the stories fall flat.
To be clear — I fully support moderation for real harm. But this isn’t that. It’s blocking harmless, classic story structures — the exact kind that inspire kids to read, learn, and imagine.
What I’m Building & Who I’m Looking For
Right now, I’m:
- Building a real-time adaptive narrative API for my classes
- Integrating GPT and image generation to teach ESL through storytelling
- Managing 14 private students and pulling in about $6K USD/month as a solo educator/developer hybrid
- Looking for like-minded devs or educators to join me
I’ve spent over 10 years in China as a new teacher trainer and vice principal. I know what engages these kids. I’ve seen what AI can do in the classroom. And I have a ton of ideas waiting to be built.
I’ve started testing other tools like Grok, MidJourney, and Gemini to piece together what I originally hoped DALL·E and GPT would do natively — but I’d much rather stay inside the OpenAI ecosystem if the filters didn’t block so much.
If you’re a developer, prompt engineer, educator, or OpenAI staff — I’d love to collaborate, share ideas, or just connect.
And OpenAI, if you’re reading this: please don’t let over-censorship kill what could be one of your most powerful educational tools.
Let’s make this usable for the people who are actually out here trying to use it.
— Tyler McCallum
Tyler’s AI Powered ESL — Education Evolved