An interesting read from the front lines of MIT research on some of the bottlenecks and mistakes in AI development so far. Artificial Intelligence is a very young field, we've only recently had the computing power to run the powerful algorithms necessary for complex thought.
Processing power has been doubling at a steady rate according to Moore's Law for several decades now. This is greatly increasing AI capabilities in brute force calculations, making them superior to humans at certain narrow tasks (predicting weather patterns, for example).
But in other aspects of AI programming, in more convoluted tasks, the software and the algorithms we produce is lagging behind our rocketing hardware capabilities. The development of these programmes still depends on human creativity and work. AI is still incapable of grasping the basic subtleties of human language for example, something a human child masters reflexively.
Since we still have a very incomplete understanding of the human brain, professionals in the field are trying to tackle the problem in a plethora of interesting ways. If research teams such as this one at MIT can successfully identify misguided avenues which are bogging down progress, we can expect to see bigger leaps in the near future.
From the article:
“[H]ow do you model thought?” In AI research to date, he says, “what’s been missing is an ecology of models, a system that can solve problems in many ways,” as the mind does.
Part of this difficulty comes from the very nature of the human mind, evolved over billions of years as a complex mix of different functions and systems. “The pieces are very disparate; they’re not necessarily built in a compatible way,” Gershenfeld says. “There’s a similar pattern in AI research. There are lots of pieces that work well to solve some particular problem, and people have tried to fit everything into one of these.” Instead, he says, what’s needed are ways to “make systems made up of lots of pieces” that work together like the different elements of the mind. “Instead of searching for silver bullets, we’re looking at a range of models, trying to integrate them and aggregate them,” he says.