TOK@ISPrague

As machine learning becomes more powerful, the field’s researchers increasingly find themselves unable to account for what their algorithms know — or how they know it.

So what, exactly, did the algorithm “learn” about the process of dying? And what, in turn, can it teach oncologists? Here is the strange rub of such a deep learning system: It learns, but it cannot tell us why it has learned; it assigns probabilities, but it cannot easily express the reasoning behind the assignment. Like a child who learns to ride a bicycle by trial and error and, asked to articulate the rules that enable bicycle riding, simply shrugs her shoulders and sails away, the algorithm looks vacantly at us when we ask, “Why?”

Kids discover lying as early as age 2, studies have found. In one experiment, children were asked not to peek at a toy hidden behind them while the researcher withdrew from the room under false pretenses. Minutes later, the researcher returned and asked the child if he or she peeked.

The tattoo stretched across the man's chest. It said "Do Not Resuscitate." His signature was tattooed at the end. "We were shocked," remembers Holt. "We didn't know what to do." The tattoo, and the hospital's decision about what it required of them, has set off a conversation among doctors and medical ethicists around the country about how to express one's end-of-life wishes effectively, and how policymakers can make it easier.

What makes one human better at talking about odors than another? Are English speakers flawed smellers, or are the hunter-gatherers of the Malay Peninsula exceptional? Advertisement Continue reading the main story The answer might come down partly to culture, suggests a study published Thursday in Current Biology.

Does bad behavior in movies or other media lead people to behave badly? There’s plenty of research on the link between onscreen media and risky behaviors like unprotected sex, binge drinking, fast driving and even violence. One large meta-analysis of such studies concluded that exposure to risk-glorifying media is associated with risky behaviors by people who consume that media. But causality issues plague most studies in this area: People who engage in risky behaviors may prefer to consume risk-glorifying media. These studies also tend to measure attitudes in controlled lab settings rather than in real life.

When the artist Peggy Weil first learned about the National Ice Core Laboratory, a few years ago, she was captivated. She contacted Geoffrey Hargreaves, the lab’s curator, and soon found herself inside a giant freezer, bundled in an Arctic-ready parka. (The temperature was minus thirty-eight degrees Fahrenheit.) With the help of lab assistants, she loaded up a cart with cannisters made of thick cardboard, each containing a small segment of a two-mile-long core from the Greenland ice sheet. Weil trundled her specimens to a cylindrical scanner and photographed them in high resolution. Eventually, she strung together eighty-eight scans, top to bottom. Then she animated them and added an accompanying score, creating a four-and-a-half-hour video, designed to be projected onto a wall.

The thought experiment is meant to show how an optimization algorithm, even if designed with no malicious intent, could ultimately destroy the world.

Thousands of lives were prolonged or saved every year because patients declared brain-dead—a form of death eventually adopted by the United Kingdom, Canada, Australia, and most of Europe—were now eligible to donate their organs. The philosopher Peter Singer described it as “a concept so desirable in its consequences that it is unthinkable to give up, and so shaky on its foundations that it can scarcely be supported.” The new death was “an ethical choice masquerading as a medical fact,” he wrote.