Developing a Universal Verbal Medium
Fostering Collaboration with AI: Kaiming He's Journey at MIT
Over the past decade, there's been a massive transformative shift, particularly since 2012, with the onset of the "deep learning revolution." Kaiming He, now a professor at MIT's Schwarzman College of Computing, attributes this shift to the power of machine-learning methods based on neural networks, making it possible to tackle various problems across different areas.
"While AI was initially seen as a high wall, it's now acting as a bridge, bringing together various scientific disciplines," He explains.
He joined MIT in 2024 as the Douglas Ross (1954) Career Development Professor of Software Technology in the Department of Electrical Engineering and Computer Science. “I’ve found myself conversing with professors from different departments nearly every day, discussing deep learning, machine learning, neural network models — even though I don't fully understand their specific area of research,” he reflects.
This cross-disciplinary collaboration is essential: from using video analysis to predict weather and climate trends to hastening the research cycle and reducing expenses in new drug discovery.
One noteworthy example of this collision between AI and other scientific disciplines is AlphaFold, an AI program developed by Google DeepMind. AlphaFold, a completely different scientific discipline, leverages the same AI tools to predict protein structures[1][2].
By lowering the barriers between scientific disciplines, AI encourages researchers to explore new solutions and uncharted paths. Remarkably, many modern AI tools, like neural networks and diffusion models, have their roots in earlier scientific fields[1][2].
He highlights the symbiotic relationship between AI and science: "Researchers in various disciplines equip us with new challenges and problems, helping us to refine and improve these tools. In return, AI's powerful methods can aid these scientists in their discoveries."
In short, the fusion of AI and science promises a new era of collaboration and technological advancement. "MIT," He insists, "is at the forefront of this shift. With its focus on interdisciplinary work and bridging different domains, MIT fosters an environment where scientists and technologists from diverse backgrounds can exchange ideas and push the boundaries of human knowledge."
While this symphony of disciplines is just beginning to play, we can expect it to become the standard within a decade or less, as AI becomes a fundamental tool for scientists across the board. "Just as everyone uses computers nowadays," He concludes, "soon, everyone will be integrating AI into their research practices, making it a part of their fundamental language for problem-solving."
References
- AlphaFold: learnt protein structure determination combined with experimental data
- AlphaFold Database
- NSF Institute for Artificial Intelligence and Fundamental Interactions
- AI, Machine Learning, and Healthcare
- The Role of AI in Cross-Disciplinary Research
- Kaiming He emphasizes that the power of machine-learning methods based on neural networks has facilitated tackling a variety of problems in different areas, including chemistry, biology, physics, and engineering.
- The collaborative efforts between AI researchers and professors from diverse scientific departments at MIT, such as electrical engineering, chemistry, and biology, are essential for predicting weather and climate trends, expediting new drug discovery, and more.
- One significant example of AI's impact on a different scientific discipline is AlphaFold, an AI program developed by Google DeepMind, which predicts protein structures in biology.
- AI has lowered the barriers between scientific disciplines, encouraging researchers to explore new solutions and uncharted paths, such as artificial-intelligence-driven advancements in the field of energy.
- Through cross-disciplinary collaboration, AI tools like neural networks and diffusion models have found new applications in various scientific fields, including robotics and environmental science.
- By refining and improving AI tools, scientists from different disciplines, like physics and engineering, help further their potential and support discoveries in their respective fields.
- In memory of the symbiotic relationship between AI and science, Kaiming He emphasizes that MIT, focused on interdisciplinary work and bridging different domains, is at the forefront of the shift toward AI integration in research practices.
- In the future, as AI becomes an essential tool for scientists across various disciplines, such as computing, engineering, and climate science, everyone will integrate AI into their research practices, making it a part of their fundamental language for problem-solving.
