Group items tagged

My father would have been unable to “teach to the test.” He once complained about errors in a sixth-grade math textbook, so he had the class learn math by designing a spaceship. My father would have been spat out by today’s test-driven educational regime.

A career in computer science makes you see the world in its terms. You start to see money as a form of information display instead of as a store of value. Money flows are the computational output of a lot of people planning, promising, evaluating, hedging and scheming, and those behaviors start to look like a set of algorithms. You start to see the weather as a computer processing bits tweaked by the sun, and gravity as a cosmic calculation that keeps events in time and space consistent. This way of seeing is becoming ever more common as people have experiences with computers. While it has its glorious moments, the computational perspective can at times be uniquely unromantic. Nothing kills music for me as much as having some algorithm calculate what music I will want to hear. That seems to miss the whole point. Inventing your musical taste is the point, isn’t it? Bringing computers into the middle of that is like paying someone to program a robot to have sex on your behalf so you don’t have to. And yet it seems we benefit from shining an objectifying digital light to disinfect our funky, lying selves once in a while. It’s heartless to have music chosen by digital algorithms. But at least there are fewer people held hostage to the tastes of bad radio D.J.’s than there once were. The trick is being ambidextrous, holding one hand to the heart while counting on the digits of the other.

The future of education in the digital age will be determined by our judgment of which aspects of the information we pass between generations can be represented in computers at all. If we try to represent something digitally when we actually can’t, we kill the romance and make some aspect of the human condition newly bland and absurd. If we romanticize information that shouldn’t be shielded from harsh calculations, we’ll suffer bad teachers and D.J.’s and their wares.

As information overload grows ever worse, computers may become our only hope for handling a growing deluge of documents. And it may become routine to rely on a machine to analyze and paraphrase articles, research papers, and other text for you.

Parsing language remains one of the grand challenges of artificial intelligence (see “AI’s Language Problem”). But it’s a challenge with enormous commercial potential. Even limited linguistic intelligence—the ability to parse spoken or written queries, and to respond in more sophisticated and coherent ways—could transform personal computing. In many specialist fields—like medicine, scientific research, and law—condensing information and extracting insights could have huge commercial benefits.

The system experiments in order to generate summaries of its own using a process called reinforcement learning. Inspired by the way animals seem to learn, this involves providing positive feedback for actions that lead toward a particular objective. Reinforcement learning has been used to train computers to do impressive new things, like playing complex games or controlling robots (see “10 Breakthrough Technologies 2017: Reinforcement Learning”). Those working on conversational interfaces are increasingly now looking at reinforcement learning as a way to improve their systems.