The TableAI Journal · Original Series
Examining how artificial intelligence misunderstands Black language—and how communities can help shape and share in the value of better technology.
Read the first story Explore the series ↓The first story
THE PROBLEM · OCTOBER 27, 2026 · 6 MIN READ
AAVE is a systematic variety of English. When technology mishears it—or makes assumptions about the person using it—the responsibility goes far beyond the speaker.
Read Part 1Grounded in research
Mishearing a speaker and judging a writer by dialect are different problems. Both deserve scrutiny.
2020 · PNAS · Speech recognition
35%Black speakers
19%White speakers
Average word error rates across five commercial speech-recognition systems in Koenecke and colleagues’ study.
Historical findings from the study sample; not a measurement of today’s products.
Read the study2024 · Nature · Dialect prejudice
Hofmann and colleagues found negative stereotypes and discriminatory judgments in hypothetical scenarios among the language models tested.
Controlled research scenarios; not measured outcomes in deployed hiring or justice systems.
Read the studyThe complete editorial series
Start with the problem. Follow the research. Explore what ownership and economic participation could look like.
The problem
Why language technology's failures with AAVE affect more than convenience.
The research
What the speech-recognition and dialect-prejudice studies actually establish.
The root cause
Training coverage, annotation quality, linguistic expertise, and evaluation design.
Real-world consequences
The stakes for customer support, hiring, education, healthcare, and beyond.
The economics
The value of annotation, cultural knowledge, and data licensing.
Ownership
Consent, compensation, stewardship, reuse, and community governance.
The solution
Linguist-led annotation, regional representation, and measurable improvement.
The vision
A vision for paid expertise and meaningful community economic participation.
From the archive
Better visibility in AI-generated answers begins with structured, trusted organizational knowledge.
A seat at the table
Bring your language expertise to the conversation, or talk with us about evaluating how your AI handles AAVE and cultural context.
Explore the specialist network Discuss an evaluation