About Me
Abhinav Madahar (अभिनव ਮਦਾਹਰ) is an independent computer scientist. His research focuses on advancing the reasoning capabilities of frontier artificial intelligence models. He also contributes to artificial intelligence law and policy discourse.
More precisely, his research asks the following:
How can we structure reasoning so that architectures which use language models can solve problems too complex for a language model to solve itself?
How can we do so while supplying a human-scale amount of time and compute, e.g. when an airline decides whether to offer a new route?
How can we do so while supplying a humanity-scale amount of time and compute, e.g. if a WHO-backed multilateral commission pours one hundred trillion USD, approximately humanity’s annual economic output, to ask an AI to cure cancer for humanity?
