A Blog by Jonathan Low

 

Apr 16, 2017

The Ways AI Will Redefine Human Intelligence

The bigger question may be how AI will change our perceptions of what human capabilities could - or should - be. JL

Adrienne LaFrance reports in The Atlantic:

As machines advance and as programs learn to do things that were once only accomplished by people, what will it mean to be human? Artificial intelligence will prove that carving out any realm of behavior as unique to humans—like language—is wrong. Disney is exploring the use of interactive humanoid robots at its theme parks, according to a patent filing. Machine progress will change our perceptions of human abilities.
The machines are getting smarter. They can now recognize us, carry on conversations, and perceive complex details about the world around them. This is just the beginning.
As computers become more human-like, many worry that robots and algorithms will displace people. And they are right to. But just as crucial is the question of how machine progress will change our perceptions of human abilities.
Once a job can be done by a computer, it changes the way people think about the nature of that job. Let me give you an example. Travel with me, if you will, back to the year 1985. Here we are in Baltimore, Maryland. It’s Christmastime.
Hutzler’s Department Store is all twinkle lights and glass ornaments. Somewhere among the festive plaid tablecloths and polished silver are Tinsel and Beau—two full-blown animatronic talking reindeer. Tinsel’s frosted in glitter and Beau’s in a top hat. I’m one of the children lined up to greet them. Baltimoreans may remember Hutzler’s as a beautiful old-fashioned department store, once known for its extravagant window displays and ornate façade. (The flagship store closed in 1989.) It was celebrated for its traditions, but also for its innovations. Hutzler’s installed the city’s first escalator in the 1930s, a modern convenience that it touted in advertisements.
So it made sense that Hutzler’s would also have such impressive animatronic reindeer. I thought of them recently when I was watching a video demonstration of Handle, the new Boston Dynamics robot that can wheel around with alarming swiftness and jump four feet into the air. Boston Dynamics also made a robot reindeer once—and, well, this is how you fall into an internet rabbit hole. Next thing I know, I’m searching the web for evidence of a faint childhood memory: a pair of beloved robotic reindeer from Baltimore in the 1980s.
It didn’t take long.
“Tinsel and Beau are back!” said one 1983 advertisement in The Baltimore Sun. “Our talking deer make such a charming couple, and what wonderful conversationalists they are!”
What I learned next, however, was that Tinsel and Beau weren’t robotic at all. A classified ad that Hutzler’s placed in The Baltimore Sun in 1986 offered a “fun opportunity for drama or theater students to be the voices of our famous talking reindeer.”
Tinsel and Beau were people!
Which, I mean, of course they were people. It seems silly now that I ever thought otherwise. Department stores in the 1980s didn’t just go around buying high-tech robotic reindeer that could carry on lengthy conversations with little kids. We’re talking about an era when Teddy Ruxpin—a furry tape cassette housed in the body of a mechanical bear—was considered a technological marvel. If you wanted to have a chat with a mobile device, your best bet was to get a Speak ‘n’ Spell to burp out the alphabet in your direction.
And yet my misplaced memory of the reindeer is understandable, maybe, given how dramatically the world has changed in the past 35 years. It’s the same reason that people today are surprised to learn that R2-D2, the lovable whistling droid from the Star Wars franchise, was operated by a human actor. (Today, another actor operates the unit for some shots; while a radio-controlled device is used for others, according to The Guardian.)
In a world of digital assistants and computer-generated imagery, the expectation is that computers do all kinds of work for humans. The result of which, some have argued, is a dulling of the senses. “The miraculous has become the norm,” Jonathan Romney wrote in an essay about computer-generated imagery for Aeon. “Such a surfeit of wonders may be de-sensitizing, but it’s also eroding our ability to dream at the movies.”Our ability to dream, elsewhere in the arts, may be intact, but computers are encroaching on all sorts of creative territory. There are computers that can forge famous paintings with astounding accuracy—and there are algorithms designed to identify such fakes. Artificial intelligences can already write novels, and there’s at least one literary contest—the Hoshi Shinichi Literary Award—that’s open to non-human competitors. Computers can flirt. They can write jokes. (Not great ones, but hey.)
Computers are now so pervasive that we should expect them to be everywhere. The past is quickly becoming a place where the presence of humans, talking reindeer and otherwise, is now surprising. That’s likely to continue, and to expand into our most creative spaces. “The unresolved questions about machine art are, first, what its potential is and, second, whether—irrespective of the quality of the work produced—it can truly be described as ‘creative’ or ‘imaginative,’” Martin Gayford wrote in an essay for MIT Technology Review last year. “These are problems, profound and fascinating, that take us deep into the mysteries of human art-making.”
As machines advance and as programs learn to do things that were once only accomplished by people, what will it mean to be human?
Over time, artificial intelligence will likely prove that carving out any realm of behavior as unique to humans—like language, a classic example—is ultimately wrong. If Tinsel and Beau were still around today, they might be powered by a digital assistant, after all. In fact, it’d be a littler weird if they weren’t, wouldn’t it? Consider the fact that Disney is exploring the use of interactive humanoid robots at its theme parks, according to a patent filing last week.
Technological history proves that what seems novel today can quickly become the norm, until one day you look back surprised at the memory of a job done by a human rather than a machine. By teaching machines what we know, we are training them to be like us. This is good for humanity in so many ways. But we may still occasionally long for the days before machines could imagine the future alongside us.

0 comments:

Post a Comment