LAKE WALES, Fla. — IBM is investing $240 million in joint development work with the Massachusetts Institute of Technology to take artificial intelligence to the next level in humanlike capabilities over the next 10 years. The new MIT-IBM Watson AI Laboratory, collocated with IBM's Watson Health and IBM Security headquarters at Kendall Square (near the MIT campus in Cambridge, Mass.), will gather more than 100 AI experts to advance AI capabilities, physical architectures, and applications, particularly to expand Watson’s expertise in health care and cybersecurity.
Beyond advancing the state of the art in current, proven AI algorithms, such as deep learning, the Watson AI Lab will look to create algorithms that mimic other critical brain functions. Concurrently with its technical work, the research effort will evaluate the economic and ethical implications of AI for society, with an emphasis on maximizing AI’s positive impact on health care and cybersecurity.
MIT president L. Rafael Reif (left) and John E. Kelly III, senior vice president of IBM Cognitive Solutions and IBM Research, shake on the 10-year partnership.
To get started, the Watson AI Lab will concentrate on three gaping needs that surpass the almost exclusive concentration of current researchers on deep-learning algorithms derived from work done in the 1980s and 1990s. First, more-advanced AI algorithms will develop and expand the capabilities of AI based on recent discoveries about parts of the working brain beyond the cortex, where deep learning occurs. For example, humans can come up with ad hoc solutions for problems of any complexity by leveraging continuous learning, which effects radical changes in the brain's structure that allow nonlinear extrapolation. Watson AI Lab researchers will use the new findings about the human brain to create brain algorithms that obsolete old ones for tackling, for instance, Big Data conundrums, no matter how complex.
AI is already pinpointing cancer and even schizophrenia symptoms in human brains.
Second, the Watson AI Lab will aim to redefine the physical materials, subsystems, and overall architectures of e-brains — especially those that leverage the analog functions of the brain — to maximize training speed; ease deployment; and incorporate newly minted technologies, such as quantum computing. The lab intends to invent brainlike quantum devices and incorporate them into a new breed of analog algorithms that achieve quantum speeds with far less hardware than would be required for analog or digital processors used alone.
IBM is combining ‘wet lab’ experiments with computational chemistry tools to create new materials for machine learning.
Third, in tandem with the work at the Watson AI lab, the IBM Watson Health and IBM Security headquarters in Kendall Square will develop biomedical and cybersecurity applications that leverage the new AI techniques. The researchers will explore AI solutions for ensuring the privacy of medical data and personalization of health care, including optimal treatment plans for specific patients, in conjunction with health care delivery to a broader range of people, nations, and enterprises. The hope is that AI can level the playing field for the delivery of both medical and cybersecurity solutions, so that individuals will not have to the foot the full bill for the costly services of modern medical and cyber experts.
The fruits of the lab will be split evenly between open-source material that fosters the ethical application of AI to everyone who can benefit from it and private-sector endeavors that encourage MIT faculty and students to commercialize the lab’s inventions and innovations.
IBM and MIT researchers working at the new lab will apply AI to cybersecurity.
The creation of the Watson AI Lab builds on a 2016 pact between IBM and MIT's Department of Brain And Cognitive Sciences. IBM also has a five-year, $50 million effort under way with MIT and Harvard University to research AI and genomics. The Watson AI Lab is by far the most ambitious of these initiatives.
The lab is co-chaired by Dario Gil, IBM Research vice president of AI, and Anantha P. Chandrakasan, dean of MIT’s School of Engineering. Engineers looking for more information on the lab or wishing to join its ranks should contact the Watson AI Lab directly.
— R. Colin Johnson, Advanced Technology Editor, EE Times