Have you ever wondered how artificial intelligence could reshape communication for the Deaf community? I’m Jeff Shaul, and as one of the co-founders of GoSign.AI—where I lead our technology efforts—I think about that question every day. At GoSign.AI, we’re dedicated to democratizing AI for sign language users, opening up new possibilities for accessibility and inclusion.
I’m also a member of the Coalition for Sign Language Equity in Technology (CoSET), a new organization working toward becoming a non-profit. Recently, my work with the Coalition gave us the opportunity to provide a Deaf-informed perspective to the FCC’s Consumer Advisory Committee (CAC) on their notice of proposed rulemaking (NPRM) around robocalls and the Telephone Consumer Protection Act (TCPA).
During this presentation, alongside AnnMarie Killian (TDI), Star Grieser (RID), and Tim Riker (Language Access Advocate), we emphasized the need for equity and functional equivalence in technology through responsible AI. It was an exciting chance to show how AI can be harnessed to improve access for people with disabilities, all while staying within legal boundaries. Stick around, and you’ll learn more about the key takeaways from our presentation and how we’re working to ensure AI technology evolves in a way that serves the Deaf community with fairness, accessibility, and accountability.
1. The Charge from the FCC
The FCC is exploring how AI can be used to assist individuals with disabilities in making phone calls, and they are considering making exceptions for AI-driven robocalls designed for this purpose. To ensure these efforts align with the Telephone Consumer Protection Act (TCPA), the FCC is also focused on balancing AI use with future anti-robocall initiatives. They have tasked the Consumer Advisory Committee (CAC) with producing a report that reflects consumer perspectives on the issue. Key areas of focus include:
How AI can enable individuals with disabilities to make phone calls.
Making AI-driven robocall exceptions for disability support.
Encouraging AI use while adhering to TCPA regulations.
Educating consumers on AI technologies and their applications.
2. Introducing the Interpreting SAFE AI Task Force
The Interpreting SAFE AI Task Force, established in 2023, works to develop industry guidelines for the responsible integration of AI into spoken and sign language interpreting. This task force includes stakeholders from across the AI and interpreting landscape—vendors, buyers, qualified practitioners, and end-users—focusing on the intersection of sign language interpreting with AI technology.
3. Coalition for Sign Language Equity in Technology (CoSET)
Though still a new organization with an evolving mission and vision, CoSET has already taken steps to promote responsible AI use for the Deaf community. The group has contributed to the production of resources like the #DeafSafeAI Report and SAFE AI Guidance, available through the SAFE AI Task Force website. Their efforts so far have included:
Hosting webinars, symposia, and workshops to engage and inform the community.
Providing guidelines on how to safely integrate AI technologies into interpreting and communication tools.
4. Principles of Automated Interpreting by Artificial Intelligence (AIxAI)
In our presentation to the CAC, we outlined four key principles for developing AI that serves sign language users:
End-User Autonomy: Deaf users must have control over how AI is integrated into their interactions.
Safety and Wellbeing: AI should improve safety for end-users by minimizing errors and maximizing positive outcomes.
Transparency: The general public and end-users need to trust the quality and transparency of AI systems.
Accountability: Companies must be accountable for any errors or harms that AI systems may cause.
5. Sociotechnical Systems: AI in a Complex World
We emphasized that AI is not just a technical challenge—it is a sociotechnical system. These systems blend social and technical elements, and their success relies on how well they address both. The AI systems we develop to interpret sign language are deeply influenced by the data they use, the environment they operate in, and the societal context in which they are applied.
For example, regional variations in ASL can cause significant issues if not accounted for. In Maine, for instance, one way to sign PORTLAND is different from the sign used elsewhere, where it means PURPLE. Failing to account for these regional dialects could lead to the suppression of unique cultural variations, or worse, mistranslations that confuse users.
This complexity is why AI must be trained on diverse datasets that account for such variations in language and pragmatics. It's not just about translating words—AI must recognize non-manual markers, classifiers, and other pragmatic elements of communication. Errors could arise if these nuances are not fully considered, potentially alienating users or delivering inaccurate interpretations.
6. AI and Accessibility in the Telephone Consumer Protection Act
Another critical discussion point was how AI-enabled interpreting tools fit within the broader framework of the Telephone Consumer Protection Act (TCPA). AI solutions that assist people with disabilities in making phone calls should be encouraged, but they must be carefully designed to comply with existing laws. We emphasized that exceptions for AI-driven robocalls must be crafted with care to ensure both compliance and safety for users.
7. Calls to Action
We concluded by urging the FCC to align its objectives with the needs of the Deaf community. We stressed the importance of involving Deaf stakeholders in every stage of AI design. Key strategic actions we recommended for the FCC include:
Eliminate Barriers: Work with CoSET and other like-minded organizations to establish standards for AIxAI to eliminate barriers for underserved communities.
Promote Awareness: Conduct programs and outreach that raise awareness about AI and its role in communication technologies.
Include Deaf Stakeholders: Ensure that Deaf users are included at every step of the AI design and development process.
The Coalition for Sign Language Equity in Technology (CoSET) advocates for responsible AI development that prioritizes the autonomy, safety, and wellbeing of Deaf users. By working closely with organizations like the FCC, CoSET hopes to establish a framework that ensures AI technologies are accessible, transparent, and equitable for all. The future of AI in interpreting lies in its ability to adapt to the diverse needs of sign language users, and through initiatives like the SAFE AI Task Force, this vision is becoming a reality.
Sincerely,
Jeff Shaul
For more details on the #DeafSafeAI Report and SAFE AI Guidance, visit safeaitf.org/guidance.