It started as a novelty. A robot hand signing the alphabet. An avatar slowly producing finger-spelling on a screen. Early versions of AI-powered ASL tools were more curiosity than communication. But fast forward to 2025, and the field looks very different.
Now we’re seeing tools that can mimic facial expressions, modify hand-shape dynamics, and even generate grammatically accurate ASL phrases. Google’s SignGemma, for example, is a powerful demonstration of how far we’ve come in using machine learning to create more human-like ASL renderings. It’s fast, fluid, and in many cases, eerily accurate.
So the question is rising louder than ever:
Could AI one day replace human interpreters?
As the Head of Interpreting at Gateway, and as someone who has spent years working alongside and advocating for Deaf and Hard of Hearing communities, I want to offer an honest take:
AI in ASL interpreting is getting better, but it’s not a replacement. It’s a tool. And like any tool, its impact depends on how we use it and who it serves.
The Promise: Access in More Places, More Often
Let’s start with the good news. There’s a lot of exciting work happening in this space, and some of it is being done with the Deaf community, not just for them. That distinction matters.
Projects like SignAll, SignGemma, and AI-based avatar systems are working with linguists and Deaf consultants to better capture the complexity of American Sign Language (ASL). And that’s no small feat. ASL is a rich, spatial, visual language with its own grammar, rhythm, and cultural nuances. It’s not just about hand movements, it’s about eyebrows, body shifts, eye gaze, and timing. Capturing that with code is incredibly challenging, and yet, these projects are getting impressively close.
In settings where live interpreters are unavailable, think rural clinics, late-night emergency rooms, or public information screens, AI tools have the potential to increase access for Deaf individuals who otherwise might receive no support at all. That’s meaningful. It’s not theoretical. And it shouldn’t be dismissed.
The Florida Atlantic University project, which uses machine learning to analyze 2D and 3D video of signers and generate accurate digital translations, is one more example of how researchers are pushing the envelope with real, applied linguistics in mind. If used ethically, tools like this can help bridge short-term gaps, especially in low-resource areas.
But Let’s Be Clear: Technology Doesn’t Equal Equity
With every step forward in ASL AI tools, there’s a growing temptation, especially from institutions and corporations, to look at this tech and think, “Problem solved.”
Let me say this clearly: AI is not a substitute for a human interpreter.
It may produce a string of grammatically correct signs. It may even look and move like a fluent signer. But it doesn’t understand the historical context of those watching it, and it can’t respond dynamically to a client’s confusion, to a sudden change in tone, to a shift in cultural or emotional meaning. It doesn’t think or feel, and that’s often exactly what’s needed, especially when interpreting in a hospital, courtroom, or classroom.
Language is more than information delivery. It’s a connection.
Interpreters aren’t just translating, they’re constantly navigating nuances. They’re clarifying, checking in, advocating, and holding space for communication to flow both ways.
No AI tool, no matter how sophisticated, can replace that kind of presence. And when organizations begin to see AI as a cost-saving measure rather than a complementary resource, we risk creating even more inequity, not less.
The Real Danger? Disconnection
What concerns me most about AI in interpreting isn’t the technology itself, it’s how quickly people forget that Deaf access is not just about words on a screen or signs in the air. It’s about respecting understanding and being understood. It’s about cultural competences woven within exchanges. It’s about being in relationship with the Deaf community, not just building tools about them.
Gallaudet University’s linguistics department released a powerful statement on this very issue, emphasizing that while AI tools can support learning and access, they must never become a reason to underfund or undervalue human interpreters. I couldn’t agree more.
Gateway’s mission is rooted in communication access, but even more deeply, in equity. That means we’re always open to innovation, always eager to see what’s next. But we also stay grounded in the real-life experiences of the people we serve.
If Deaf individuals tell us a new tool works for them, we listen. If they say something feels off, we listen even more.
So What’s the Path Forward?
Here’s what I believe: AI is here, and it’s not going away. Nor should it. There are amazing possibilities for education, language modeling, interpreter training, and backup communication in low-access settings. These are wins.
But the path forward must be rooted in partnership with the Deaf community. Not assumptions. Not automation. And certainly not outsourcing equity to an algorithm.
At Gateway, we will always champion tools that expand communication access, when they’re ethical, inclusive, and used as part of a bigger picture. And we’ll always stand up to remind the world that human interpreters are not optional. They’re vital.
By Dave Coyne
If you want to connect to discuss this topic further, please reach out to me at dcoyne@gatewaymaryland.org to schedule a time to meet.
Learn More About Gateway
Gateway connects people to their worlds and aids individuals in their ability to understand and to be understood. Gateway has grown into an organization that serves more than 8,000 children and adults every year, helping them communicate more effectively. With programming both on our Baltimore campus and through community-based programming, we provide education, access, and medical support to anyone who needs it.
We envision a society where everyone can understand and be understood and where everyone is treated with integrity, compassion, and equity.
