[ad_1]
Q. Why are you so excited to take on what seems to be one of the most stressful and least popular jobs in academia?
A. I can’t wait to take on this role now because of what’s going on in Washington DC. Under the latest version of the Endless Borders Act, which is now part of the U.S. Innovation and Competition Act, Congress is considering estimated $ 200 billion in funding for science. and technological research. Off the charts! Just a fraction of that money could stretch Columbia in new directions. This could lead to deeper collaborations with businesses, government labs, and nonprofits, help us diversify our research labs, and expand our links with K-12 schools and local communities.
Q. You ran the Data Science Institute for four years, but your research is focused on AI. What is the difference?
A. Data science and AI overlap but are not the same thing. The goal of data science is to understand the world through data, which today means digital data. There is a cycle of life data science: generating, collecting, processing, storing and managing data; then analyze and visualize the results, and tell a story about your analysis. AI brings new techniques to the analysis phase, but the ultimate goal of AI is to build a machine with real intelligence.
Most of our advanced AI technologies today are based on deep learning algorithms ingesting huge data sets and producing models for decision making. The more data, the better the model. Deep learning models are now as good, if not better, than humans at many narrowly defined tasks. When the deep mind is AlphaGo beating the world’s best go player (who happened to be Chinese) was China’s Sputnik moment. China stood up and said, “By 2030, China will dominate the world in AI.
Q. How is AI changing the way research is done? What does this mean for Columbia?
A. In traditional computing, people write programs. In machine learning, people feed data into the computer and the computer itself writes the program; this learn the program from the data. The term machine learning is relevant here. The machine learns the rules on its own. Because the machine, and not the human, writes the program, the program is not easily interpretable for us. In the case of deep learning, the most successful machine learning technique to date, we don’t really understand the science of how it works or how successful it is. This is an example of applications ahead of theory.
These tools are already part of our daily lives. AI systems recommend movies and books, respond to our voice commands, and translate web pages from one language to another. AI is also adding to our repertoire of scientific methods. In medicine, deep learning models process medical tests faster than humans and detect warning signs that even experts sometimes miss. And they never tire of it! In astronomy, they analyze images from telescopes and space probes to make new discoveries about our universe. In climate modeling, they help reduce the uncertainty surrounding climate change and its impacts.
These tools are accelerating science, and I expect the trend to continue. AI also holds great promise for the social sciences. At Microsoft, I saw how bringing together economists and machine learning experts helped the company better forecast sales for certain products.
Q. What are you most proud to accomplish at the Data Science Institute?
Create bridges. Everything I did was aimed at establishing collaboration between schools and disciplines. The Institute of Data Science connected many points across campuses and beyond the gates of Columbia. When people from different perspectives and areas of expertise come together, sparks spring up. Thanks to data science, researchers and educators have asked questions they never thought to ask, let alone answer.
I also feel good about creating the Trustworthy AI initiative to investigate some of the unintended consequences of machine learning. Our goal is to find out if the AI systems that make decisions about people’s lives are reliable: do I really have cancer? Is the moving object in front of my car a ball or a child? Will the bank approve my loan? It turns out that it is difficult to formally define the properties of reliability, let alone prove and guarantee that an AI system has one.
A. Columbia Engineering and the Data Science Institute built the IBM Center on Blockchain and Data Transparency under your tenure. And Columbia continues to court corporate lenders. Why is industrial collaboration so vital?
In some areas of research, especially AI, the industry is ahead. They have the data, which is mostly proprietary consumer data. They also have large amounts of computing power. Amazon, Microsoft, Google have almost unlimited computing power thanks to their cloud infrastructure. They have GPU clusters that the university could never afford. I see enormous potential for collaboration. If professors could access the data and calculate, they could validate their algorithms at scale and identify new directions for research.
It is a mutually beneficial relationship. The industry is looking to academia for new ideas and new talent. Academia looks to industry for real world problems to solve and opportunities to scale solutions. It’s an important way to broaden our impact.
Q. You have held leadership positions in academia, industry and the federal government. What skills have enabled you to succeed in such different cultures?
A. Being able to listen and learn. Know what you don’t know, and surround yourself with superb talent.
[ad_2]
Source link