About $1.49M for neurodiverse middle schoolers to build AI-enabled inventions — teaching not how to use AI but where it fails (TERC)
A project developing and studying a hands-on maker program in which neurodiverse learners aged 10 to 14 design, build, program and share their own AI-enabled physical inventions using sensors, motors, lights, displays, beginner-friendly block coding and AI tools. Learners discover what AI systems can do and deliberately investigate where and why they make mistakes.
Grant overview (primary data)
- Award amount$1,492,109
- RecipientTERC Inc (Massachusetts)
- ProgramAISL
- Period2026-09-01 〜 2029-08-31
- FunderU.S. National Science Foundation (NSF) / NSF
Key points
- A project developing a hands-on maker program in which neurodiverse learners aged 10 to 14 design, build and share AI-enabled inventions using sensors and block coding.
- The core of the learning is not only what AI can do but deliberately investigating where and why it makes mistakes.
- The record states that neurodivergent young people often have strengths suited to technical design, while many AI learning experiences are language-heavy and abstract.
- Outputs include free activity guides, build instructions, facilitator supports and accessibility materials for museums, makerspaces and afterschool programs nationwide.
- The 120 NSF awards this site holds as of 2026-08-31 span 67 programs, averaging 1.8 each; the largest, CyberAICorps SFS, holds 11.
- The 120 awards divide across 67 programs, the largest, CyberAICorps SFS, at 11 — investment spread across many frameworks.
1Putting investigation of error at the centre
Most efforts to teach AI aim at letting people experience what it can do. What this project places at the centre of learning is deliberately investigating where and why it goes wrong. Run it, find where it does not work, think about why. Children follow the same sequence of testing and improvement that technical designers actually carry out.
It is a design that gives the perspective of someone who fixes AI, rather than someone who uses it, from the start.
2Who language-heavy material leaves out
The reason the record gives is specific. Neurodivergent young people often possess strengths suited to technical design — pattern recognition, systematic reasoning, creativity, persistence — yet few travel that path. One reason cited is that many AI learning experiences are language-heavy and abstract. So this project takes the form of building by hand with sensors, motors, lights and displays.
The starting point of the design is a judgment that the cause lies in how something is taught rather than in what is taught.
3120 awards spread across 67 programs
Grouping the 120 NSF awards this site holds as of 2026-08-31 by program name gives 67 distinct programs — an average of 1.8 awards each, with the largest, CyberAICorps SFS, at 11. EPSCoR follows at 8, PCL at 7, SBIR Phase II at 7 and AI Research Institutes at 6. NSF investment in AI is not concentrated in a few large plans but distributed across many different frameworks.
AISL, the program this project belongs to, supports informal STEM learning, addressing museums and afterschool settings rather than classrooms.
Why it matters
Training for AI tends toward teaching how to use it. This project puts investigating where it fails at the centre, pointing toward people who can evaluate and improve. The observation that language-heavy material leaves some learners out is a point that carries into how corporate training is designed.
FAQ
What is AISL?
Why investigate mistakes?
Why build by hand?
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: 2628608