NTT Research has established a new Physics of Artificial Intelligence Group at Harvard, focusing on the complexities of AI and its decision-making processes.
The group aims to address the 'black box problem' in AI, which relates to the lack of understanding of how AI systems make connections and decisions.
Led by Dr. Hidenori Tanaka, the group merges physics, psychology, philosophy, and neuroscience to explore AI's mysteries and improve trustworthiness and safety in AI systems.
The Physics of Artificial Intelligence Group is a spin-off of NTT Research's PHI Lab, with a focus on understanding the neural network-like structures of AI.
Dr. Tanaka emphasizes the importance of defining concepts like creativity and kindness in mathematical terms for AI systems to comprehend.
The group will collaborate with Harvard University Center for Brain Science and Stanford University to advance research in AI mechanisms and human-AI fusion.
By adopting a cross-industry approach and integrating physics into AI research, NTT aims to design more cohesive and trustworthy AI technologies.
The group's interdisciplinary exploration of AI reflects a historical tradition of understanding technology and human relationships in physics.
Dr. Tanaka sees AI as a neural network akin to the human brain, highlighting the need to apply physics principles to study AI's inner workings.
The group's mission is to spark conversations and learning about AI across various backgrounds, contributing to the advancement of AI research and development.