MIT moves to probe human and artificial intelligence - Financial Times

1/2/2018

MIT moves to probe human and artificial intelligence

Academics seek to regain initiative on machine learning ceded to tech industry

The IQ initiative is not the first MIT programme to explore links between human and machine intelligence

Clive Cookson in London

Massachusetts Institute of Technology is launching an ambitious programme, involving hundreds of researchers across the university, to understand human intelligence and apply that knowledge to develop intelligent machines.

The MIT Intelligence Quest or MIT IQ, based at an institution that has been at the forefront of artificial intelligence research since the 1950s, is a far-reaching academic effort to regain the initiative in AI. It comes at a time when the technology industry’s fast-growing research labs threaten to suck the field’s most talented scientists, engineers and ideas away from universities.

“We plan to raise hundreds of millions of dollars from the private and public sectors for this effort,” said Anantha Chandrakasan, MIT dean of engineering. “We are having discussions with a broad set of companies that have shown great enthusiasm for the initiative.”

MIT professors say tech companies have acknowledged concerns that they are poaching too many AI experts from universities, where they are needed to teach future scientists and engineers as well as carry out basic research. The industry is responding by increasing academic collaboration and funding.

“Two weeks ago we had a symposium with several Google and Alphabet companies,” said Josh Tenenbaum, professor of cognitive science and computation. “We can see that industry is keen to replenish the well of ideas and people.”

MIT hopes to sign collaborations similar to that announced in September with IBM, in which the computing company agreed to invest $240m in a new MIT-IBM Watson AI Lab.

We had a symposium with several Google and Alphabet companies. We can see that industry is keen to replenish the well of ideas and people

Prof Josh Tenenbaum

The MIT IQ programme sets out to answer two big questions, said Rafael Reif, the university’s president: “How does human intelligence work, in engineering terms? And how can we use that deep grasp of human intelligence to build wiser and more useful machines?”

It will aim to deliver the answers through two linked entities. The first, called “The Core”, will work on the science and engineering of human and machine intelligence in general. As well as gaining fundamental understanding of how natural and computer brains work, it will generate machine-learning algorithms for more specific applications.

The second entity, “The Bridge” will apply discoveries in natural and artificial intelligence to a wide variety of disciplines including disease diagnosis, drug discovery, materials and manufacturing design, automation, synthetic biology and finance. Both will collaborate with research labs worldwide.

MIT already has more than 200 “principal investigators” (senior scientists) working on aspects of natural and artificial intelligence, said Prof Reif. “We are amplifying this research by providing additional resources and connecting researchers more closely together.”

Harnessing technology to challenge inequality

When AI becomes too big to fail

Both robots and Big Tech have to earn our trust

Prof Tenenbaum said he was looking forward to reproducing in a machine the way human intelligence develops from birth though infancy and childhood. His colleague Dina Katabi, professor of computer science, wants to build an intelligent home for patients with chronic disease, which would constantly monitor their health and predict problems and forestall emergencies before they occur.

Adrian Weller, senior research fellow at the Alan Turing Institute in London, the UK national research centre for AI, said: “We warmly welcome this MIT initiative. AI has made great progress in the last few years but we are still very far from enabling machines to do things that our human brains can do without effort.”

He added that he was particularly pleased that, like the Turing Institute, MIT would investigate the social and ethical implications of AI alongside its scientific and technological development. “If we are going to apply AI across society, we need to think hard about the ethical issues.”