When parents teach their children to connect with the world, they teach through associations and patterns. Take the letter s as an example. Parents show their children enough examples of letters that soon after, they will be able to identify other examples with inactive guidance; schools, a book, billboards.
The emerging artificial intelligence (AI) technology is teach Same way. The researchers provided the right example for the system, they hoped it could be identified, and like young children, AI began to recognize patterns and push this knowledge into an environment that had never been seen before, forming its own “neural network” for classification. But, like human intelligence, experts lose their opinion on understanding AI decisions.
Therefore, the “black box problem” of AI emerges because we do not fully understand how or why AI systems make connections, nor do we understand the variables in their decisions. This issue is especially important when seeking to improve the credibility and security of systems and determine the governance adopted by AI.
From AI-powered vehicles that fail to brake and hurt pedestrians in time to AI-biased health technology devices that can help doctors diagnose patients, and the biases shown by AI-hiring screening processes, the complexity of these systems has led to the rise of new areas of research: the physics of AI, aimed at further building AI to further build tools for AI, thus making AIS ASE AN GUALSAS ASS ASS INSEAS FEAMASS INSTER INSE AN GUARES ASS INS ATE AS AN GUALSANS INSE AN EXANS INSEAL INSEALS INSEALS INSEALS INSEAL ASS INSEAL INS ATE ASS INSE INSE ASS INSE-ASS INSE-ASS INSE-ASS INSE-ASS INSE promotion.
Now, a new independent research team will address these challenges by integrating the fields of physics, psychology, philosophy and neuroscience into interdisciplinary exploration of the mysteries of AI.
News’s AI team is a derivative of the NTT Research Physics and Informatics (PHI) lab and was unveiled last week at the NTT Upgrade 2025 conference in San Francisco, California. It will continue to advance the physics understanding of AI in AI methods, which the team has been studying for the past five years.
Dr. Hidenori Tanaka, PhD in Applied PhD in Applied Physics and Computer Science and Engineering at Harvard University, will lead a new research team, based on his previous experience in NTT’s Intelligent Systems Group and CBS-NTT’s AI research program.
“As a physicist, I’m excited about the subject of intelligence because mathematically, how do you think of concepts of creativity? Do you even think of goodwill? If it’s not suitable for AI, these concepts will still be abstract. It’s easy to speculate that this is my definition of kindness, which is mathematically illustrated, but our language is certain because it’s an important thing because it can make AI important because it can make AI important. What kind of good is mathematics? yesFor example, Dr. Tanaka told me at the intervals of the upgrade meeting last week.
Early in the research, the PHI lab recognized the importance of understanding the “black box” nature of AI and machine learning to develop new systems with increased energy efficiency computing. However, advances in AI over the past five years have caused increasingly important security and trustworthiness considerations and are therefore critical to the industry application and governance decisions adopted by AI.
Through the new research team, NTT research will address similarities between biology and artificial intelligence, thus revealing the complexity of AI mechanisms and establishing a more harmonious integration of human collaboration.
Although novel in the integration of AI, this approach is nothing new. Physicists have tried for centuries to reveal the exact details of technology and human relationships, from Galileo Galilei’s knowledge of how objects move and their contribution to mechanics to steam engines’ knowledge of thermodynamics during the industrial revolution. However, in the 21st century, scientists are trying to understand how AI works in training, accumulating knowledge and making decisions so that more cohesive, secure and trustworthy AI technologies can be designed in the future.
“AI is a network of neurons that are structured in a very similar way to how the human brain works; through synaptic-related neurons, all represented by numbers inside the computer. Then, we believe physics can… physics is about taking anything from the universe, making mathematical assumptions about their inner work, and testing them, and testing them.”
The new group will continue to work with the Harvard Center for Brain Science (CBS) and plans to work with Suya Ganguli, associate professor at Stanford University, with whom Dr. Tanaka has co-authored several papers.
However, Dr. Tanaka stressed that natural science and cross-border approaches would be fundamental. In 2017, when he was a doctoral candidate at Harvard University, researchers realized he wanted to do more than traditional physics and followed in the footsteps of his predecessors, from Galileo to Newton and Einstein to open up the new conceptual world of physics.
“At the moment, AI is a topic that I can talk to everyone. As a researcher, it’s great because everyone is always talking about AI, and I learn from every conversation because I realize how people view and use AI, even beyond academic background. I think the mission of NTT is to catalyze these catalysis, and whether it’s for people’s background, we trigger these people’s backgrounds, because our backgrounds are no matter how everyone’s interactions,’’
