How AI will play a role in media accessibility

To me, it feels like this year has made one thing clear: AI is a useful tool that can help us make media even more accessible.

Thanks to AI tools enabling accessible pricing and quicker turnaround times, media accessibility is now more than within reach for organisations that hadn’t made it a priority previously.

As a project manager at a multimedia accessibility company, I have seen more interest in AI services in 2025 than in any year to date.

And as I explore in this blog, I see it as a helpful extension for humans to make media even more accessible to more people, as opposed to a direct replacement of humans.

An industry on the rise

The global AI market in media and entertainment is set to grow from $17.1 billion (in 2023) to $195.7 billion in 2033, via Market Us.

Additionally, from data revealing the leading drivers to adopting GenAI in media and entertainment according to industry business decision makers worldwide from Statista (as of September 2024), there are some interesting insights.

Insights include the opportunity to accelerate content/ media production (61%), opportunity to reduce production costs (55%), opportunity to improve viewer/customer experience (51%) and innovation (49%).

What an industry with AI actually looks like

It’s all well and good saying that the AI market is set to grow exponentially.

But what does that actually look like in the world of media production?

Well, media accessibility spans:

• Video content
• Audio content
• Live events
• Live shows
• Webinars
• Films
• Global conferences
• Community events
• Internal meetings

And here, AI can integrate with the likes of subtitles, live captions and voiceovers with the use of technology.

Does AI not pose risks?

While AI is quicker and cheaper than human workflows, there are obviously some risks that come with it.

For example, AI voiceovers will do a good job, but they will struggle with tone and emotion. Over time I think the voices will sound more natural and human-like, of course, as the training on models improves.

And for live captioning and subtitles, the accuracy won’t be as high as humans right now (we estimate between 70% – 85% accuracy on average). But again, AI is improving year on year.

Effectively, you get what you pay for, and I think clients know that. So, it’s always worth considering what the AI solution is being used for and balancing against what quality is needed.

Humans are Batman, AI is Robin

In my opinion, things work best with humans at the core of our production. But AI is like a helpful sidekick that can save the day when called upon.

By having both as a service in the media and entertainment industry, you can cover all your clients’ needs, whether they ask for AI or a human solution, or perhaps even a mixed approach.

If you had a smaller budget and wanted a mass number of voiceovers for short, explainer videos on social media, then AI would very work well (that being said, AI voiceover quality does depend a little bit on the accents and audio quality).

On the other hand, if you had an advert and you wanted to make an audience feel something, human voiceovers with personality, tone and emotion excel here.

As for subtitles and live captions, if you want a cheaper solution and a quicker turnaround time for scale on subtitles, AI sounds like it could be for you.

But if you want as close to complete accuracy as you can get (around 98% for live captions and 99.99% for subtitles), opt for humans.

AI and compliance

While AI can’t help with everything to do with compliance – like a trusty assistant – it can help with some compliance tasks.

For example, the European Accessibility Act came into play in June 2025 and it requires certain businesses (if they meets the requirements) to make its products and services accessible for everyone.

Let’s say that you have a really limited budget and need to scale subtitles to meet EAA requirements, then AI can come in handy here.

Yes, human subtitles would be ideal for optimal accuracy. But not everyone has the budget for that, so the way we see it is that accessibility is a sliding scale. Being more accessible than not is always better, even if it’s not 100% perfect.

The alternative to non-compliance is to risk fines, penalties, reputational damage and more.

Also, the Purple Pound (the spending power of disabled people and their households) is worth around £274 million in the UK alone.

So, beyond the financial risks of penalties and the like, you are also cutting off a huge potential audience who, if catered to by your company, would really value accessibility measures.

I can’t see AI in media accessibility going away anytime soon. In fact, quite the opposite, I think it will only get stronger and end up being a bigger help to humans.

That is the key though, by no means are we saying it is replacing humans in the industry, it’s just able to offer an alternative service for a different type of client to make accessibility more widely available.

As we see it, humans will still be the main character, and AI will be the trusty sidekick.