Research knowledge base · Augmentation, governance, and harm
AI and Neurodiversity
How AI systems are reshaping work, learning, and care for neurodivergent people — and the governance research that should shape design.
Overview
The current state of research
Artificial intelligence is reshaping work, learning, and care for neurodivergent people — sometimes as augmentation, sometimes as harm. Governance frameworks from the EU and U.S. NIST provide common ground, and a growing body of disability-rights research documents both the upside and the algorithmic risk.
Why this matters
For the people doing the work
Individuals
Generative AI can lower barriers in communication, executive function, and learning.
Employers
Algorithmic management tools can systematically disadvantage neurodivergent staff.
Policymakers
Governance frameworks need disability-specific provisions.
Researchers
Longitudinal outcome data on AI augmentation are early but growing.
What the research consistently shows
Findings that replicate across the literature
- Generative AI shows clear augmentation potential for executive function, communication, and learning.
- Productivity-monitoring and hiring algorithms can amplify ableism when unaudited.
- Trustworthy-AI principles (oversight, fairness, accountability) increasingly converge across regions.
- Inclusion is more often a governance problem than a technical one.
Current evidence
Organized by theme, not by citation list
Augmentation
AI scaffolding for writing, planning, communication, and learning.
Harms
Algorithmic management, biased hiring tools, surveillance-driven performance metrics.
Governance
EU HLEG, NIST AI RMF, sector-specific guidance.
Design
Co-design with neurodivergent users improves outcomes substantially.
Evidence snapshot
Quick research facts
3
Curated studies summarized
0
Systematic reviews / meta-analyses
2
Government / clinical guidance
0
Public health sources
Counts reflect the resources curated in this collection plus the synthesis above. The knowledge base is reviewed quarterly and expanded as new peer-reviewed work is published.
Key takeaways
Practical, evidence-informed conclusions
Trustworthiness is a system property, not a feature.
Governance starts with mapping the system, not buying a tool.
AI without guardrails amplifies ableism.
Common misconceptions
Myths addressed by the research
Myth
AI is neutral.
What research shows
AI systems reflect their training data, design choices, and deployment context.
Myth
Compliance equals ethics.
What research shows
Compliance is necessary but not sufficient for trustworthy AI.
AI Companion questions
Take this further
Try one of these questions in the AI Companion to keep exploring:
- How can I use AI ethically to support executive function?
- What does the NIST AI Risk Management Framework require?
- How do hiring algorithms disadvantage neurodivergent candidates?
References
Curated scholarly sources
We summarize each source in original language rather than reproducing copyrighted text. Visit each link for the underlying publication.
- Framework / consensusEuropean Commission · 2019
Ethics Guidelines for Trustworthy AI
European Commission HLEG-AI
The original EU framework for trustworthy AI — human oversight, technical robustness, fairness, accountability, and societal well-being.
Takeaway: Trustworthiness is a system property, not a feature.
View source - Framework / consensusNIST · 2023
NIST AI Risk Management Framework 1.0
U.S. NIST
Voluntary U.S. framework for managing AI risk across govern, map, measure, manage functions.
Takeaway: Governance starts with mapping the system, not buying a tool.
View source - Peer-reviewed studyAI Now Institute · 2019
Algorithmic management and disability
Whittaker, M., et al. (AI Now Institute)
Documents how productivity-monitoring and hiring algorithms systematically disadvantage disabled and neurodivergent workers.
Takeaway: AI without guardrails amplifies ableism.
View source
Related collections
Keep exploring
Continue exploring