Although cognitive and neuropsychological approaches emphasize the losses with age that might impair social perception, motivational theories indicate that there may be some gains or qualitative changes. Charles and Carstensen review a considerable body of evidence indicating that, as people get older, they tend to prioritize close social relationships, focus more on achieving emotional well-being, and attend more to positive emotional information while ignoring negative information. These changing motivational goals in old age have implications for attention to and processing of social cues from the environment. Of particular importance in considering emotional changes in old age is the presence of a positivity bias: that is, a tendency to notice, attend to, and remember more positive compared to negative information. The role of life experience in social skills also indicates that older adults might show gains in some aspects of social perception.
6-1
Up until the late 1970s, the United States produced at least 70 percent of the clothing that Americans purchased. And - thanks to the New Deal - for much of the twentieth century, brands and manufacturers were expected to follow strict national labor laws. But in the late 1980s, a new section of the clothing business cropped up: "fast fashion," the production of trendy, inexpensive clothes in vast amounts at lightning speed in subcontracted factories, to be sold in thousands of chain stores. To keep the prices low, fast-fashion brands cut manufacturing costs - and the cheapest labor was available in the world's poorest countries. Offshoring caught on across the industry, just as globalization was unfurling. Though it started as a small corner of the business, fast fashion's remarkable success was so enviable it soon reset the rhythm for how clothing - from luxury to athletic wear - was and is conceived, advertised, and sold. The impact was dramatic: in the last thirty years, fashion has grown from a $500 billion trade, primarily domestically produced, to a $2.4-trillion-a-year global giant.
6-2
The fact that deep learning computers merely carry out programmatic functions without understanding what they are doing or any implications has created problems in the past. In particular, many analysts have described a problem of algorithmic bias. The data sets that neural networks train on are representations of the world as it is rather than the world that we might like to see. Because of this, a deep learning algorithm may reproduce the worst stereotypes and biases of our society as a whole. A human, on the other hand, might exercise judgment and question the patterns that he or she was seeing. For example, a deep learning network might see that most of the pictures labeled as "secretary" in an online archive are female, whereas most pictures of a "boss" are male, and conclude that men are always bosses and women are always secretaries.
6-3
Assumptions are a reflection of what's going on in one's mind, where we focus our thinking. When our thoughts are voiced to someone else, we now have a great opportunity to learn the person's perception. It is precisely from that perception that change efforts can start. Suppose Alex was thinking, "I'm too old to get that job. They're looking for someone fresh out of school." As a result, Alex did not apply for the teaching position, even though he really wanted it. Suppose he was a great leader who could turn that school around, but his lack of initiative created a lose-lose for him and for the school district. What if instead he recognized he had unique experiences from his prior roles that could indeed be of true value? If he had been willing to challenge his own assumption, he might have taken action, with confidence, and pursued his goal. Stories like this happen too often. They hinder progress and inhibit organizations from achieving their goals. They cause individuals to remain locked in their self-imposed boundaries.
6-4
Physicians claim that a lot of what they do is intuitive. It is reasoning through the associations built up over years of practice. But when AI scientists work on the problem of medical diagnosis, their effort is to see the diagnostic process as a set of explicit procedures that can be captured in a program. AI experts attack the problem by interviewing a physician over the course of months, trying to pin down every aspect of how he or she makes decisions. They model the structure of that practical knowledge which "feels intuitive." The resulting program will, given the same information as the physician, usually come to the same conclusion. The process of writing such programs has a side effect. If the program "thinks" henceforth like the physician, the physician's thinking about his or her activity has been changed by collaboration in the making of the program. What once seemed intuitive to the physician has been shown to be formalizable.