This past Tuesday, I braved Chicago traffic to attend a talk at Venue610 that was sponsored by Chicago Ideas. I rarely attend talks, but when I heard the title, Future of Intelligence: Human, Machine and Extraterrestrial, I had to make the effort. I’m glad I did.
One of the overarching questions addressed during the presentations was whether we are engineering our own destruction via technology, whether by developing a machine intelligence capable of deciding humans are no longer necessary and destroying us, or by contacting an extraterrestrial civilization that might do the same. After an introduction by the moderator, Professor Marcelo Gleiser of Dartmouth College, came a brief presentation by Patricia Churchland, Professor of Philosophy at UC-San Diego. Her conclusions would likely alleviate concerns that a Matrix-like world is just around the corner. There is more to intelligence than single-task pattern recognition, and current neural nets just aren’t yet where they need to be to tackle anything more complicated, and may not be for the foreseeable future (5-10 years).
After Dr. Churchland, Doctor Jill Tarter of the SETI Institute took the stage to discuss the search for extraterrestrial technological civilizations. To me, the most intriguing point she discussed was that a sufficiently advanced civilization may be indistinguishable from nature, perhaps with the extremes smoothed out. Monitoring would reveal a lack of extreme weather, perhaps modification of their star.
The talk concluded with a discussion moderated by Dr. Gleiser during which he asked questions, then took questions from the audience. One point that was made with respect to extraterrestrial intelligence was that an old civilization would have had to shed aggression to sustain development—they couldn’t get old without becoming kinder and gentler. This is in contrast to the thoughts of Hawking and others who feel such a civilization would not be friendly. A point with respect to machine intelligence was that if even a small change is made to a process, a neural net may need millions of trials to relearn that altered process, while a biological system such as a dog or rat can incorporate a change in two trials. We still don’t know how memories are recovered or how neurons “talk” to one another, so we still have a long way to go before we develop an AI that could rival a biological system, much less subsume or eradicate it.
As it happens, the entire talk is available online. I recommend giving it a listen not just because it’s an interesting discussion of some very fundamental questions along with the added nuance and side comments that are missing from my brief synopsis, but also because hey, story fodder.
And also because, with all that’s been going on lately, it may be a relief to consider that the latest model computerized refrigerator may not be capable of taking over the world after all.
Not yet, anyway.
::cue computerized voice singing “Daisy”::