Many things have changed during the 15 years since Kaiming, he was a doctoral student.
“When you are in your doctoral stage, there is a high wall between different disciplines and subjects, and there was even a wall raised in computer science,” he said. “The guy sitting next to me could do things that I couldn't understand completely.”
During the seven months since he joined the Schwarzman College of Computing Schwarzman as a professor of career technology at Douglas Ross (1954) in the department of electrical engineering and computer science, he says that he experiences something which, in his opinion, is “very rare in human scientific history” – a drop in walls that exposes in different scientific disciplines.
“There is no way to understand physics, chemistry or the border of research in biology, but now we see something that can help us break these walls,” he says, “and it is the creation of a common language that has been found in AI.”
Build the AI bridge
According to him, this change began in 2012 in the wake of the “Deep Learning Revolution”, a point where it was realized that this set of automatic learning methods based on neural networks was so powerful that it could be put to later.
“At this stage, computer vision – helping computers to see and perceive the world as if they were human beings – began to grow very quickly, because it turns out that you can apply this same methodology to many different problems and many different fields,” he said. “Thus, the community of computer vision quickly became very large because these different sub-themes were now able to speak a common language and share a common set of tools.”
From there, he says that the trend has started to extend to other fields of computer science, including natural language processing, speech recognition and robotics, creating chatgpt bases and other progress towards general artificial intelligence (AG).
“All this has happened in the last decade, leading us to a new emerging trend that I really look forward to, and it is that the methodology of the AI propagates other scientific disciplines,” he said.
One of the most famous examples, he says, is Alphafold, an artificial intelligence program developed by Google Deepmind, which makes predictions of the protein structure.
“It is a very different scientific discipline, a very different problem, but people also use the same set of AI tools, the same methodology to solve these problems,” he says, “and I think that is only the beginning.”
The future of AI in science
Since his arrival in February 2025, he said that he spoke to teachers in almost all departments. Some days, he finds himself in conversation with two or more teachers of very different backgrounds.
“I certainly do not fully understand their field of research, but they will simply introduce a little context, then we can start talking about in-depth learning, automatic learning, (and) neural networks models in their problems,” he said. “In this sense, these AI tools are like a common language between these scientific fields: automatic learning tools” translate “their terminology and their concepts in terms that I can understand, then I can learn their problems and share my experience, and sometimes offer solutions or opportunities to explore.”
Expanding to different scientific disciplines has significant potential, from the use of video analysis to predict weather and climatic trends to accelerate the research cycle and reduce costs in the discovery of new drugs.
While AI tools offer a clear advantage to the work of HE scientific colleagues, he also notes the reciprocal effect they may have and have had, on the creation and advancement of AI.
“Scientists offer new problems and challenges that help us continue to develop these tools,” he explains. “But it is also important to remember that many AI tools today come from previous scientific fields – for example, artificial neural networks have been inspired by biological observations; The diffusion models for the generation of images were motivated from the physical term. ”
“Science and AI are not isolated subjects. We are approaching the same goal from different angles, and now we meet. ”
And what better place to meet it.
“It is not surprising that put it could see this change earlier than many other places,” he said. “(MIT Schwarzman College of Computing) has created an environment that connects different people and allows them to sit together, speak together, work together, exchange their ideas, while speaking the same language – and I see that it is starting to happen.”
Regarding the moment when the walls will be completely down, it notes that it is a long -term investment which will not occur overnight.
“Decades ago, computers were considered high-tech and you needed specific knowledge to understand them, but now everyone uses a computer,” he said. “I expect that in 10 years or more, everyone will use a kind of AI in one way or another for their research – these are only their basic tools, their basic language, and they can use AI to solve their problems.”
