Research publications
Research reports from HUTE.
Literature reviews and proposed frameworks. Each states its evidence status at the top, and every reference has been verified against Crossref or arXiv.
These reports are the long-form record of our thinking. None of them presents original experimental results. The first two map the evidence and propose how to measure it; the third analyses the role of human expertise in AI development. Read the methodology for how we label evidence, select sources and handle corrections.
For shorter explainers on the same topics, see Insights. To suggest a correction or to discuss a report, contact the team.
Research reports
The State of Social Intelligence in Artificial Intelligence: Capabilities, Limitations, and Open Problems
A literature review of social intelligence in AI: theory of mind, emotion, pragmatics, multi-agent behavior, evaluation limits, culture, safety risks and open problems.
Toward a Framework for Evaluating Human Understanding in AI Systems
A proposed framework for evaluating whether AI understands people: construct validity, psychology-grounded tasks, culture, expert grading and benchmark integrity.
Human Expertise as Infrastructure for the Next Generation of Artificial Intelligence
How human expertise enters AI: SFT, preferences, RL, rubrics, production traces, provenance, consent, privacy and contributor pay. A literature review.