Easy Read Pictures: The AI Nightmare That Keeps Getting Worse

If AI ever woke up in a cold sweat from a bad dream, it’d be over Easy Read pictures. Seriously, if you think AI-generated transcription has its limitations, wait until you try to get it to handle imagery. What looks like a promising tool at first glance quickly becomes a maddening exercise in frustration—and sometimes outright hilarity.

The Laughable Suggestions Problem

Let’s start with the basics. AI might be able to pair sentences with images, but the suggestions it comes up with? Often laughably unsuitable.

For example, you might ask an AI to generate a picture for a sentence like “You can contact your local council to apply for housing support.” What do you get? A random image of a shiny corporate skyscraper, or worse, a surreal, dream-like montage of office workers smiling awkwardly at a phone. Neither of these makes sense to someone trying to understand how to apply for housing. AI lacks the nuanced understanding of what context-appropriate, relatable imagery looks like for someone in the Easy Read target audience.

And don’t even get me started on abstract concepts. If you need a visual to represent “eligibility” or “support services,” you’re doomed. AI will churn out vague nonsense that confuses more than it clarifies.

One of the images generated by Shutterstock’s Design Assistant when asked to create an image for ‘support services’

The Tweak Trap

Here’s where things go from bad to worse. Let’s say you actually get an AI-generated picture that’s close to what you need—maybe even almost perfect. You’d think, Great, just a few tweaks and we’re good to go. Right? Wrong. AI doesn’t do “tweaks.” You can’t just nudge the elements into place or fix that one weird detail.

Instead, the AI spits out an entirely new picture, often with zero resemblance to the one you wanted to adjust. It’s infuriating. And as if that weren’t bad enough, the new image has a frustratingly good chance of being worse. Maybe it has even more off-kilter elements. Maybe the style suddenly changes, or some crucial piece disappears entirely. Whatever the case, you’re left stuck in an endless loop of generating and rejecting images, wishing you’d just sketched something by hand instead.

Why This Is a Dealbreaker for Easy Read

Easy Read imagery isn’t just about adding some eye-catching visuals to break up the text. The pictures play a critical role. They’re there to reinforce understanding and to bridge gaps for people who struggle with text alone. When the pictures miss the mark—or worse, actively confuse the reader—the whole purpose of Easy Read is undermined.

For example:

• If the image doesn’t directly relate to the sentence, it’s useless.

• If the image is overly complicated, it can be just as overwhelming as the original text.

• If the image is too abstract or stylistically inconsistent, it breaks the flow of understanding.

This is where AI really struggles. It doesn’t understand the why behind the imagery choices. It doesn’t know what’s going to resonate with the audience. And it definitely doesn’t have the insight to say, This picture could be clearer if we added X or removed Y.

Why the Co-Production Process Matters Even More for Pictures

Let’s bring this back to co-production because, honestly, this is where the panel of Experts by Experience really shines. When you involve them in reviewing and suggesting pictures, they’ll spot things that no AI—or even the most well-meaning professional—ever could. They’ll tell you:

• What makes a picture confusing.

• What kind of imagery feels relatable.

• Whether the image reinforces the text or creates unnecessary questions.

It’s a collaborative, iterative process. You don’t just slap an image onto a page and call it a day. The panel reviews each picture alongside the text, ensuring that it works as part of the overall communication. They might suggest tweaks, entirely new ideas, or even reject the need for a picture if it’s not helpful. This level of insight simply isn’t possible with AI.

AI’s Role in Easy Read Imagery: The Harsh Reality

Here’s the truth: AI isn’t ready for Easy Read pictures. It might get better someday, but right now, its limitations are glaring. If you want to use it for anything, you’re better off treating it as a brainstorming tool—something to generate rough ideas that can be refined (and heavily edited) by humans. But even then, you’ll need to weigh whether the time you spend wrestling with AI might be better spent just working with a designer or co-production panel directly.

The Bottom Line

As much as I’d love to say AI is the future of Easy Read, the reality is that it’s nowhere near ready to take on the most challenging parts of the process—especially when it comes to imagery. Transcription? It has potential. Pictures? Forget it. And honestly, that’s fine. Easy Read was always meant to be a collaborative, human-centered process. The value comes from the people who bring their lived experience to the table, not from shortcuts or quick fixes.

So for now, let’s keep AI in its lane and focus on what really matters: co-production, collaboration, and creating materials that actually work for the people who need them most. Easy Read deserves nothing less.