When AI Looks Impressive, But Accessibility Is Still at Risk

Yesterday, I saw media coverage about AI tools translating ASL in real time. At first, it looked impressive. I can understand why many people would see this as exciting progress.

Two sign language professionals stand side by side against a dark blue background. Both wear formal business clothing and ornate Venetian-style masks in white and gold that cover their faces. Their hands are raised in signing positions, creating a strong contrast between human communication and hidden identity. The image feels dramatic, symbolic, and slightly unsettling, suggesting themes of visibility, trust, and mystery in communication.

But after reading a public LinkedIn post by Craig Radford about the SLxAI conference in Boston, I started to think more carefully about what was being presented and why it matters. His post raised concerns about how AI signing avatars were being shown to the public and how easily that could be misunderstood. That concern makes sense to me.

This is not about rejecting technology. Innovation is important. New tools can be useful. AI may have a place in some situations.

But we also need honesty.

The real concern is not only that AI signing avatars now exist. The bigger concern is how they are being presented. Recent media coverage showed AI tools that translate between ASL and English using a digital avatar, presenting them as a way to help bridge communication gaps.

The risk is that many people may see this and assume AI avatars are now equal to human interpreters.

They are not.

Human interpreting is not only about changing words from one language into another. It involves meaning, context, tone, emotion, culture, timing, and human understanding. Real communication is not always simple or predictable. People interrupt. Conversations change direction. Meaning depends on the situation.

This is where skilled human interpreters are essential.

An AI system may be able to follow a speaker in some situations. But human interpreters follow the full discussion. They understand the flow of conversation, the people involved, the purpose of the message, and the deeper meaning behind the words.

That difference is very important.

This is why media coverage like this can be risky. A polished demonstration may look convincing on screen, but it can also create the wrong impression. People who do not understand the complexity of interpretation may think the problem is already solved. They may believe AI is now good enough to replace human interpreters.

That is dangerous.

There are also wider concerns about consent, trust, quality, and accountability. Deaf experts and community members have raised questions about whether this kind of technology can truly reflect the facial expression, nuance, and human connection needed for real interpreting in everyday life.

If businesses, organisations, schools, hospitals, or public services begin to think AI avatars are a cheaper alternative, accessibility could be reduced to appearance rather than real quality. Something may look accessible without actually providing the accuracy, trust, nuance, and human connection that true access requires.

That should concern all of us.

The SLxAI Summit itself brought together conversations about sign language, interpreting, accessibility, and AI. That is exactly why this discussion matters so much. When accessibility and AI are discussed in public, there must be clarity about what technology can do today and what it still cannot do.

This is also why I keep coming back to something I wrote earlier in my own blog: AI may be able to follow a speaker, but human interpreters follow the discussion. In my earlier post, I argued that live communication depends on context, pace, and unpredictability in ways that make human interpreters essential. I still believe that.

If we are serious about accessibility, then we must be honest about what technology can and cannot do today. AI tools may have some limited and supportive uses. But that is very different from presenting them in a way that encourages the public to think they are equal to human interpreters.

They are not.

Accessibility should never be defined only by speed, cost, or what looks impressive in a news story. It should be defined by whether people truly understand and are truly included.

We should welcome innovation, but we should also question hype.

We should support progress, but we should also protect truth.

And most importantly, we must make sure that the future of accessibility does not come at the expense of the Deaf community it claims to support.

Related links

Craig Radford’s LinkedIn post:https://www.linkedin.com/posts/craigjradford_i-just-attended-the-slxai-conference-in-boston-share-7451123086969729024-EKbv/

NBC Boston media coverage:https://www.nbcboston.com/video/on-air/as-seen-on/ai-tool-translates-asl-in-real-time/3935023/

SLxAI Summit 2026:https://www.slxai.org/2026

My earlier blog: AI Can Follow a Speaker. Human Interpreters Follow the Discussion.https://www.timscannell.co.uk/post/ai-can-follow-a-speaker-human-interpreters-follow-the-discussion