ARIA, an institute for trustworthy "AI assistants," starting from mental health — NSF AI grant $4M (Brown)
The NSF awarded about $4M to ARIA, which researches next-generation "AI assistants" starting from mental and behavioral health — a field where trust, empathy, and personalization are critical. It treats human and machine cognition as complementary, organized around three pillars: Grounding, Instructability, and Alignment.
Grant overview (primary data)
- Award amount$4,000,000 / Est. total $20,000,000
- RecipientBrown University (Rhode Island)
- ProgramAI Research Institutes, NSF-Capital One Partnership
- Period2025-10-01 〜 2030-09-30
- FunderU.S. National Science Foundation (NSF) / NSF
Key points
- Researches trustworthy next-generation "AI assistants," starting from mental/behavioral health
- Treats human cognition and machine cognition as complementary scientific endeavors
- Three pillars: Grounding, Instructability, and Alignment
- CS, neuro/cognitive science, philosophy, law, education + practitioners; K-12 to postgraduate training
- About $4M, led by Brown, NSF AI Research Institutes (Capital One partnership), 2025–2030
- Of the 120 NSF AI-related awards this site holds as of 2026-09-01, close to nine in ten run to a whole number of years, led by 1,095 days at 47.
The NSF awarded about $4,000,000 to the AI Research Institute on Interaction for AI Assistants (ARIA), led by Brown University (NSF Award 2433429; program: AI Research Institutes / NSF-Capital One partnership; October 2025 – September 2030).
1Fields where trust and empathy are decisive
Per the abstract, ARIA accelerates the development of next-generation AI assistants for mental and behavioral health — a field where trust, empathy, and personalization are critical and where current AI systems fall short. Its distinctive stance is to treat human cognition and machine cognition as inherently complementary scientific endeavors.
It brings together researchers in computer science, neuroscience, cognitive science, philosophy, law, and education with mental-health practitioners and civil-society groups, advancing technology while improving human well-being. It grows a future-ready workforce through interdisciplinary education from K-12 through postgraduate training.
2Three research pillars
ARIA's research centers on three interconnected pillars. (1) Grounding: new models for efficient learning and generalization, learning algorithms yielding rich causal models, and new evaluation metrics for trustworthy AI assistants. (2) Instructability: new paradigms for establishing trust in AI, theories and models of how humans interact with AI, and methods for describing AI's internal processing.
(3) Alignment: advancing human-centered design, precisely defining what it means to be aligned, and developing metrics of alignment in complex ethical and social contexts. Throughout, it promotes integration between academia and industry and between research and continuing education.
3Periods are set in whole years
Counting how many of the 120 NSF AI-related awards this site holds as of 2026-09-01 share the same number of days, 1,095 days — three years — leads with 47, then 1,460 days with 28, 1,825 days with 17 and 730 days with seven.
That is 104 awards, close to nine in ten, running to a whole number of years. Research awards are built on annual appropriations, so the term follows the year. Unlike procurement contracts, which include forms one day short at 364 or 729 days, an award takes the calendar year straight into its design.
Why it matters
A large U.S. investment studying "AI assistants" together with human cognitive science from the angles of trust, empathy, and alignment. A useful read for those tracking conversational-AI/agent reliability and safety, and healthcare AI.
FAQ
Why start from mental health?
What is AI "alignment"?
Sources (primary)
Source: NSF Award Search (U.S. National Science Foundation, public domain). Amounts are the obligated amount. For privacy, we do not handle principal investigator names.
- NSF Award (original, official)
- NSF Award ID: 2433429