Biographical excerpt on Danny Hillis and early AI research

modern computer design. “Language,” Boole wrote, “is an instrument of human reason, and not merely a medium for the expression of thought.”187 Hillis has a magnetic intellectual charisma, as you might have guessed by now. An afternoon with him resembles nothing so much as a lingering mental theme park: Roller coasters of big ideas (a 10,000 year clock!) mixed with smaller sugary treats (how to design a better fencepost). No wonder he fit in so well at Disney. Critics accused Steve Jobs of having a “reality distortion field”, in which the Apple founder’s charisma bludgeoned the boundaries of the practical. Hillis, by contrast, has a sort of “reality enhancement field” in which much of the world as seen through his eyes or heard in his light-hearted voice is sharper, filled with possibility. From an early age Hillis had been interested in the dream of a thinking robot. Maybe it was that the constant uprooting of his childhood left him with a giddy sense that it was easier to assemble your own friends than to try to make them new at each stop. But somehow this led him to the idea of an artificial brain, which was Danny’s main idea when he arrived at MIT in the fall of 1972. The tinker-toy tic-tac-toe computer he built was a nod to this hope, but it’s jerry-rig aesthetic masked deeper ambitions. “Someday, perhaps soon, we will build a machine that will be able to perform the functions of a human mind,” Hillis wrote at the start of his PhD thesis a few years later. “A thinking machine.”188 What Hillis and others like his mentor Marvin Minsky, realized was that the human brain works differently than machine logic. Life, after all, is not a series of linear math problems. (Much as we might wish it was at times.) You look outside. It occurs to you to say to your wife, “What a lovely day.” This is not a result of some “a then b then c” calculation, but rather the product of thousands of simultaneous inputs and twitches dancing through the space of your consciousness. If you were to process that same thought in a linear fashion, it might look like this: First, look at the sky, examine the cloud-to-blue ratio, check for too much wind, sense the temperature, open mouth. Your wife would be out the door before you'd even begun to speak. The ability to operate on many different pieces of data all at once is one of the most striking, enviable features of the human mind. But of course, you've probably, recognized: That is fundamentally a network problem. How do you act instantly, everywhere? Connection. So it was that, a dozen years after his tic-tac-toe machine, Hillis began work on a device designed to think super fast. Faster than any computer ever had. He called it The Connection Machine. “The ability to configure the topology of the machine to match the topology of the problem will turn out to be one of the most important features of the Connection Machine,” he wrote. Adding, in case the 187 “Language’: George Boole, An Investigation of the Laws of Thought: On Which Are Founded the Mathematical Theories of Logic and Probabilities. (New York, N.Y.: Dover Publications, 1951) 188 “Someday”: William Daniel Hillis, The Connection Machine, PhD. Thesis Submission, MIT (1985) p. 2 133 HOUSE_OVERSIGHT_018365