How intelligence works — and how to extend it.
NOVARC organizes its work around six research directions and a single conviction: technology should strengthen the mind, not quietly erode it. The evidence is what makes that urgent.
A pattern is emerging.
Sustained interaction with generative AI is associated with measurable changes in how people think — and the findings are converging across institutions.
Your Brain on ChatGPT
Kosmyna et al.
Participants who used LLMs to assist with writing showed reduced neural connectivity, weaker memory of their own work, and a diminished sense of authorship over the output they produced.
“Cognitive debt — capability lost in exchange for short-term efficiency.”Read study ↗
Generative AI and Critical Thinking
Lee et al.
Higher confidence in AI was associated with reduced critical thinking; higher self-confidence in one's own ability was associated with more of it.
“The more people trust the system, the less they exercise the faculties it is meant to support.”Read study ↗
Reasoning is the infrastructure every functioning system depends on. When it weakens at scale, the cost is mistaken for normal— until it isn't.
NOVARC was founded to address that question with the seriousness it requires.
Any technology that interfaces with human cognition should be required to demonstrate that it strengthens — rather than diminishes — the faculties it engages. The same standard we apply to medicine, education, and physical training.
Built on an argument from cognitive science.
The human brain learns by being challenged, not by being served. Three findings shape everything we build.
The brain learns through prediction error.
The brain's learning signal is the gap between what we expected and what actually happened. When reality matches our prediction, no learning occurs. When it doesn't, the brain updates.
Today's AI minimizes that gap by design. Ask a question, get an answer. No surprise, no learning.
Productive failure beats direct instruction.
Learners who struggle with a problem before being taught the solution develop deeper conceptual understanding than those given the solution upfront.
ChatGPT removes the struggle. We build it back in.
Calibration is trainable — and most people lack it.
Humans are systematically overconfident, and overconfidence correlates with worse learning. Targeted calibration training reliably improves metacognitive accuracy.
AI that always agrees produces users who feel more confident while becoming less able.
Six planets. One sun.
Each direction orbits the same objective. Explore them — every planet is one of the six.