Good AI Content

Paul Meyer, deaf since birth, relied on interpreters and captioners for communication throughout his HR career.

With the pandemic’s rise in video conferencing, companies started using AI transcription, assuming it could replace human interpreters.

Problem

AI captioning is often inaccurate, especially in noisy environments or with non-standard accents.

Companies unaware of these limitations began using AI without consulting disabled employees, leaving them at a disadvantage.

Example

Meyer experienced misunderstandings due to AI inaccuracies, which led to job challenges.

He believes his lack of access to effective communication contributed to being let go when the company downsized.

Progress and Efforts

Tech companies are working to improve voice recognition for diverse speech.

Google’s Project Relate app aims to help users with non-standard speech, including those with a deaf accent.

Deaf-led startups like OmniBridge are developing tools to bridge gaps in informal conversations.

Concerns

Assumptions of Accuracy: Employers often assume AI captioning is flawless, which can create communication gaps.

Limited Deaf Representation: Deaf voices are often missing in AI development, leading to tools that don’t fully meet their needs.

Lower Standards in Tech for Disabled Users: Some tech products for the Deaf community lack quality due to biases or lack of consultation with Deaf people.

Lessons Learned

AI Isn’t a Substitute for Human Interpreters: Human understanding remains essential, especially for complex or nuanced conversations.

Inclusive Development is Key: Involving disabled people in product development ensures tools better meet their needs.

Quality Standards Matter: Companies must prioritize high-quality, accessible tools, or risk leaving disabled employees behind.

Advocacy and Awareness: Deaf employees and allies must raise awareness about AI’s limits to ensure fair accommodations.

Conclusion

For true inclusivity, companies need both technological and human solutions and must commit to understanding and addressing the limitations of AI-driven accessibility tools.