Teaching Robots to Imagine

Can robots be taught to imagine? Google’s DeepMind artificial intelligence group is doing just that –– developing computer versions of what many consider humanity’s quintessential trait. The software world long has pursued sentient consciousness as its holy grail. But until now, it’s only been found in science fiction movies like A.I., Ex Machina, and Transcendence. DeepMind engineers say they have cracked the code by combining two kinds of machine-learning. The first is linear, which is nothing new, with the computer applying a predefined algorithm over-and-over till it finds answers and then remembering them. In the second more radical approach, the computer tries many algorithms to find which work best, and then changes the very way it approaches problems. Combining the purely linear with a more systemic approach, DeepMind’s “Imagination-Augmented Agent” mimics intuitive learning in a way prior software hasn’t. It’s not exactlythe same as human imagination, but it comes closer than ever before to what neuroscientists say the brain does.

While robotic imagination may be improving, human thought isn’t faring as well. Most people feel uncreative and without inspiration, as discussed in earlier chapters. Corporations say innovation is withering as well. Novelist Ursula Le Guin recently observed that, “In America today imagination is generally looked on as something that might be useful when the TV is out of order. Poetry and plays have no relation to practical politics. Novels are for students, housewives, and other people who don’t work.”[i]Beyond the abandonment of a creative genre or two, American society also is undergoing a wholesale commodification of imagination itself. Disney is most famous for this, its “Imagineering” (imagination + engineering) brand one of the most viciously protected anywhere. But hundreds of companies evoke imagination to conjure an aura of specialness ––seen in promotions like Bombay Safire’s “Infused with Imagination,” GE’s “Imagination at Work,” Electrolux’s “Power to Capture Imagination,” Lego’s “Imagine,” Microsoft’s “Imagine Academy,” Nestle’s “Feed your Imagination,” Samsung’s “Imagine,” and Sony’s “Made of Imagination.” Continue reading “Teaching Robots to Imagine”

Intelligence community tries to predict what’s next

Spying is all about predictions: about knowing what someone else can or will do next, about thinking how to win.

At least that is how governments tend to think about spying. In this context it makes sense that the Obama administration would spend some time prognosticating. Turns out they spend billions to satisfy their curiosity, as the U.S. and many other countries have done for decades. Asia Times carries a story today about this curious and expensive enterprise, which begins with the paragraphs below:

“Think of it as a simple formula: if you’ve been hired (and paid handsomely) to protect what is, you’re going to be congenitally ill-equipped to imagine what might be.imgresAnd yet the urge not just to know the contours of the future but to plant the Stars and Stripes in that future has had the US Intelligence Community (IC) in its grip since the mid-1990s.

“That was the moment when it first occurred to some in Washington that US power might be capable of controlling just about everything worth the bother globally for, if not an eternity, then long enough to make the future American property. Continue reading “Intelligence community tries to predict what’s next”