OPINION:
Sounds so easy — teaching machines common sense. After all, artificial intelligence can already drive cars, recommend musical choices, pinpoint locations on maps and even help identify crimes before they occur. What’s the big deal about A.I. that can think through the “whys” of the moment and consider the “wherefores” of what’s to come before offering up the next best course of action — or inaction, as the case may be?
Plenty, as it turns out.
When it comes to building artificial intelligence with good old-fashioned common sense, elusive is thy name. Many have tried. Many have failed.
DARPA, the Defense Advanced Research Projects Agency, aims to rectify that by bridging technology with — get this — psychology.
The logic?
Just this: that the key toward creating machines that can properly judge is to go back to the days of toddler-time — to go back and study how human minds progress from the purely instinctive and real-time reactive of baby days to the judging, perceiving and understanding ways of more adult maturity. So says DARPA, anyway.
This is the future of A.I. — machinery that doesn’t just do but understand and reflect accordingly.
“The absence of common sense prevents an intelligence system from understanding its world, communicating naturally with people, behaving reasonably in unforeseen situations, and learning from new experiences,” said Dave Gunning, a program manager for DARPA’s Information Innovation Office, in a statement. “This absence is perhaps the most significant barrier between the narrowly focused A.I. applications that we have today and the more general A.I. applications we would like to create in the future.”
It’s the difference between, say, A.I.-fueled crime-tracking cameras red-flagging on every pedestrian deviance that occurs on a monitored street — and A.I. instead recognizing, without specific coded instructions, that the man rushing down the sidewalk and reaching in his chest pocket is late for work, fumbling for his phone while chasing down a taxi. Nothing nefarious here; nothing to see, move on, no need to waste time on a false positive.
“During the first few years of life, humans acquire the fundamental building blocks of intelligence and common sense,” Gunning said. “Developmental psychologists have found ways to map these cognitive capabilities across the developmental stages of a human’s early life, providing researchers with a set of targets and a strategy to mimic for development a new foundation for machine common sense.”
Interesting. And useful to training technology, apparently.
DARPA’s reaching out to researchers in both applicable fields, psychology and A.I., to do what hasn’t yet been done — to bring to the world a new level of intelligent machinery that doesn’t just take orders, but also, in layman’s of layman’s, assesses orders, as well.
The potential for technological advancement is magnificent. Think of all the strides from thinking-like-man machinery that could come for national security and the military — all the applications that would follow for the fields of health care, law enforcement, human resources and more. It’s mind-boggling. Eye-widening. But the flip side? The downer? The elephant in the room?
Let’s hope this machinery never outpaces its human creators. There’s no telling where technology, powered by super-intelligence combined with common sense judging, would go — and not necessarily under the leadership of mankind.
• Cheryl Chumley can be reached at cchumley@washingtontimes.com or on Twitter, @ckchumley.
Please read our comment policy before commenting.