AGI research
ActiveWe study general intelligence that reasons well with limited data, limited compute, and human values.
Focus areas
Where we focus.
Alignment beyond feedback tuning.
Values a system holds for reasons it can explain, not only behavior it was rewarded for.
Cognitive architectures.
Bayesian, causal, and symbolic reasoning, combined with deep learning.
Intelligence under limits.
Doing more with less data and less compute.
Inspired by nature
Brains learn from little data and run on very little power. We study bio-inspired cognitive architectures (BICA) for ideas that machines can borrow, and compare them with mathematical models of intelligence.
Read the comparison: Bio-inspired vs. mathematical AGI: BICA and AIXI compared.
We also study philosophical frameworks, including Vedic traditions, for what they say about knowing, attention, and judgment, and what they might teach general intelligence.
PhD students and postdocs are welcome. See careers.
Collaborate
Research with us.
Professors, labs, and companies can propose a joint project. Tell us what you want to study together.
Choose AGI as the program.