efsli's stance on AI and sign language interpreting: our vision for the future

The European Forum of Sign Language Interpreters (efsli) recently took part in an important two-day event in Brussels on April 7–8, 2025: "Sign Language in the Age of Artificial Intelligence: What Digital Rights and Ethical Dilemmas?" Organized by the European Union of the Deaf (EUD), this gathering brought together experts, Deaf community leaders, researchers, and professionals from all over Europe. We came together to understand what Artificial Intelligence (AI) means for sign language, accessibility, and the rights of Deaf people.

AI isn't neutral: why we need a human touch

One major point became incredibly clear: AI is not neutral. As sign language interpreters, we're already seeing how new AI tools—like computer-generated avatars, speech-to-text systems, and machine translation—are changing our field. Too often, these technologies are developed without enough input from the very people they impact: Deaf communities and professional interpreters.

When technology is built without our involvement, there are big risks. Companies might try to cut costs by replacing human interpreters, and biased AI systems could actually make communication worse, not better.

efsli's clear position: ensuring AI works FOR us

In response, efsli is taking a strong, proactive stance. We shared our initial thoughts during the event, and here's what we believe is crucial:

• Protect human interpreting: We must have laws that guarantee the right to a human interpreter, especially for essential services like healthcare, education, and legal matters.
• Include interpreters and deaf people in design: Interpreters and Deaf professionals must be involved in every step of AI development, from the first idea to the final product.
• Train future interpreters for AI: Interpreter training programs need to teach about AI, helping future interpreters understand its uses and potential pitfalls.
• Explore new roles for interpreters: AI could create new jobs for interpreters in areas like ensuring AI quality, advising on ethics, and evaluating accessibility.
• Support deaf leadership in tech: Deaf communities should lead and own the development of new sign language technologies.

Interpreting is more than just words: AI can't replicate connection

It's vital to understand that the interpreting profession isn't just threatened by AI; it's evolving. Interpreting isn't simply about swapping words. It's about connecting cultures, conveying emotions, and understanding subtle references. No computer algorithm can truly replicate the human touch—the cultural sensitivity, emotional intelligence, and ability to adapt spontaneously that human interpreters bring to complex, real-world situations. This human element is especially critical in sensitive areas like health consultations, educational settings, courtrooms, and public announcements. As efsli emphasized: "Interpreting is not just about translating words. It’s about connecting worlds, emotions and cultural references. No algorithm can do that."

Unequal treatment: a concern for all sign languages

Our partners, the European Union of the Deaf (EUD), share our concerns. They've highlighted a worrying imbalance: AI for spoken languages is usually presented as a tool to help human interpreters (for example, with transcriptions or note-taking). But for sign languages, AI is often pitched as a replacement, with signing avatars proposed for places like museums or public announcements. This often happens without involving Deaf interpreters and ignores the complex nuances of true communication.

This difference shows an unequal valuation of signed languages. As the EUD pointed out:

• AI still lacks the precision and cultural understanding needed for good sign language interpretation.
• Current automated solutions are often inaccurate, risking misinformation and excluding Deaf people.
• We need to apply the same level of caution to AI for visual languages as we do for spoken languages.

The EUD is working with European institutions to ensure AI rules protect accessibility and respect the integrity of sign languages. As interpreters, we fully support this call. It's unacceptable for our work—and the languages we interpret—to be valued less than those of our spoken language colleagues.

Linguistic justice: protecting the richness of sign languages from AI

Beyond the direct impact on interpreters, efsli is deeply concerned about linguistic justice for sign languages themselves. Many AI systems for signed languages are trained on limited or problematic data. For example, they might use recordings from formal TV interpreting, which often doesn't capture the rich, natural diversity of a country's sign language.

This reliance on narrow data can be dangerous:

• Forced Standardization: If AI only learns from limited sources, it could unintentionally push for a "standard" way of signing. This ignores regional differences, unique grammar, and the expressive variations that make each national sign language vibrant. It could slowly erase the linguistic diversity that Deaf communities have a right to.
• Loss of Authenticity: The beauty of sign languages comes from their 3D nature, how space is used, facial expressions, body language, and cultural nuances. If AI is trained on formal, "cleaned-up" interpreting, it will miss the natural flow and idioms of everyday Deaf communication. This could lead to technically correct but culturally empty interpretations.
• Built-in Bias: If the training data itself is biased (e.g., featuring only certain types of signers or formal settings), the AI will learn and repeat these biases, further marginalizing some users or ways of signing.

To ensure linguistic justice, AI for sign languages must be developed using diverse, representative, and ethically sourced data, created hand-in-hand with Deaf communities and sign language experts. We must actively fight against any technology that diminishes the natural richness and evolution of our national sign languages.

Our path forward: a collaborative future

These positions are just the start. efsli plans to delve deeper into these discussions throughout 2026–2027. Our aim is to build a common, constructive framework across Europe to understand AI's impact on our profession and, more importantly, to prepare future interpreters to thrive in this changing landscape.

We're committed to continuing this conversation with everyone involved: the interpreting community, Deaf organizations, training institutions, and policymakers. AI is here to stay. Our job now is to ensure it develops in a way that respects diverse languages, protects human dignity, and strengthens interpreting as a valuable, people-centered profession.

Stay tuned for more opportunities to join this vital conversation. We invite all national associations, interpreter trainers, and stakeholders to help us shape a future where technology supports inclusion—not the other way around.

#AIandSignLanguage #DigitalInclusion #efsli #DeafRights #InterpreterTraining #EthicalAI #AccessibilityMatters #EUD #SignLanguageEquality #LinguisticJustice