Research

MIT Quest is comprised of two pillars: the MIT Quest Core, aimed at understanding intelligence and developing intelligence technologies, and the MIT Quest Bridge, aimed at fostering the use of intelligence tools at MIT and beyond.

MIT Quest Core

The mission of the MIT Quest Core is to reverse engineer intelligence. Intelligence — what it is, how the brain produces it, and how it can be engineered — is simultaneously the greatest open question in the natural sciences and the most important engineering challenge of our time. At MIT, we believe these challenges are interlocked aspects of the same grand mission. By bringing together diverse teams to work at the intersection of science and engineering, we aim for breakthroughs in both human and artificial intelligence.

MIT Quest Bridge

The mission of the MIT Quest Bridge is to leverage basic research breakthroughs in intelligence to build novel platforms, tools, and services to apply intelligence tools broadly across MIT and beyond. The MIT Quest Bridge focuses on four paths: We will develop new tools to collect and curate the massive amounts of data required to accurately represent the physical world. We will create more powerful hardware to model bigger and more complex phenomena than is currently possible. We will build intelligent software that is intuitive and can be integrated across labs, classrooms, businesses, and beyond. We will develop a framework to deploy intelligent systems ethically and transparently.

Quest Research Affiliates

MIT–IBM Watson AI Lab

The MIT–IBM Watson AI Lab aims to advance AI hardware, software, and algorithms related to deep learning and reasoning; increase AI’s impact on industries, such as health care and cybersecurity; and explore the economic and ethical implications of AI on society. Co-led by Antonio Torralba, inaugural director of the MIT Quest, and David Cox at IBM, and chaired by Anantha Chandrakasan, Dean of the School of Engineering, and Dario Gil, vice president of AI and IBM Q at IBM Research, the Lab will invest $240 million in AI efforts over the next 10 years. Current funded projects are answering questions such as:

  • How can advanced algorithms expand capabilities in machine learning and reasoning?
  • How can quantum computing optimize machine-learning algorithms and other AI applications?
  • How can AI ensure the security and privacy of medical data?

MIT-SenseTime Alliance on Artificial Intelligence

The MIT-SenseTime Alliance on Artificial Intelligence opens new avenues of discovery across MIT in areas such as computer vision, human-intelligence-inspired algorithms, medical imaging, and robotics. SenseTime, founded by MIT alumnus Xiao’ou Tang PhD ’96, specializes in computer vision and deep learning technologies. The MIT-SenseTime Alliance will provide funding for 27 projects involving 50 principal investigators from MIT departments and labs within all five schools. Current funded projects are answering questions such as:

  • How can linguistic theory transform machine-learning algorithms to better approximate how people converse?
  • How can artificial systems, like robots, learn common sense knowledge?
  • How can tools in product design and systems architecture capitalize on strategies that combine human and machine intelligence?

Abdul Latif Jameel Clinic for Machine Learning in Health

The Abdul Latif Jameel Clinic for Machine Learning in Health (J-Clinic) harnesses the power of machine learning for preventative medicine, clinical diagnostics, and drug discovery and development. J-Clinic uses a holistic approach that draws on MIT’s expertise in cellular and medical biology, computer science, engineering, and the social sciences as it focuses on developing machine learning technologies to revolutionize the prevention, detection, and treatment of disease. It concentrates on creating and commercializing high-precision, affordable, and scalable machine learning technologies in areas of health care ranging from diagnostics to pharmaceuticals, with three main areas of focus: Preventative medicine methods and technologies with the potential to stop non-infectious disease in its tracks. Cost-effective diagnostic tests that may be able to both detect and alleviate health problems. Drug discovery and development to enable faster and cheaper discovery, development, and manufacture of new pharmaceuticals, particularly those targeted for individually customized therapies.

MIT-Liberty Mutual Insurance Collaboration

The MIT-Liberty Mutual Liberty Collaboration represents a $25 million, five-year commitment to advance artificial intelligence research in computer vision, computer language understanding, machine learning fairness, data privacy and security, and risk-aware decision making, among other topics. Research topics under discussion include efforts to make decision-making algorithms transparent to customers and regulators, use computer vision to reduce crashes by identifying dangerous driving conditions and roadways, further protect the anonymity and security of personal data, use computer language understanding to analyze insurance claims and speed processing and compensation, and structure investment portfolios.

Center for Deployable Machine Learning 

From self-driving cars to real-time language translation, technology once dismissed as science fiction is close to becoming reality. Or is it? The truth is most AI systems aren’t ready for prime time. They’re prone to failure beyond the narrow settings in which they were designed, vulnerable to malicious interference and may carry unexpectedly high societal costs when scaled. To build next generation AI systems that are robust, trustworthy and socially responsible, we need a radically different approach. The new MIT Center for Deployable Machine Learning offers the breadth and depth of knowledge to help AI achieve its potential in science, industry and society. Led by Aleksander Madry, a professor at MIT’s Department of Electrical Engineering and Computer Science, the center brings together 11 faculty experts in interpretability, security, computer vision and applied machine learning.