Submitted by Angela F

Perspective: Person with a disability

Choosing a thumbnail for a YouTube video is a visual task. And, like most visual tasks, it can be a royal pain when you’re blind.

But AI is making things easier.

When I upload a video to YouTube, I typically choose the option to select a thumbnail from the video itself. YouTube presents three possible images, usually with one suggested as the default.

Here’s where ChatGPT comes in.

I take a screenshot of the three thumbnail choices, upload it to ChatGPT, and ask it to describe each one.

Now I know what’s actually in the pictures. I can find out whether the presenter is clearly visible, whether the image looks professional, or whether somebody was caught in the middle of making a ridiculous face.

I can also ask more specific questions. For example, if Dan is the primary presenter, I might ask which thumbnail shows him wearing his famous hat. That helps me confirm I’m selecting an image featuring the right person.

Most of the time, YouTube’s suggested thumbnail turns out to be the best choice anyway. But now I can make that decision based on information rather than simply trusting the default.

And that’s the important distinction.

AI isn’t choosing the thumbnail for me. It gives me access to visual information so I can make the choice myself.

Visual tasks have historically been some of the most frustrating parts of using technology as a blind person. But with tools like ChatGPT, they’re becoming more manageable every day.

Sometimes accessibility isn’t about changing the software. Sometimes it’s about finding a new way to accomplish the task.

That’s accessibility in real life.