1 . Flinging brightly coloured objects around a screen at high speed is not what computers’ central processing units were designed for. So manufacturers of arcade machines invented the graphics-processing unit (GPU), a set of circuits to handle video games’ visuals in parallel to the work done by the central processor. The GPU’s ability to speed up complex tasks has since found wider uses: video editing, cryptocurrency mining and most recently, the training of artificial intelligence.

AI is now disrupting the industry that helped bring it into being. Every part of entertainment stands to be affected by generative AI, which digests inputs of text, image, audio or video to create new outputs of the same. But the games business will change the most, argues Andreessen Horowitz, a venture-capital (VC) firm. Games interactivity requires them to be stuffed with laboriously designed content: consider the 30 square miles of landscape or 60 hours of music in “Red Dead Redemption 2”, a recent cowboy adventure. Enlisting AI assistants to churn it out could drastically shrink timescales and budgets.

AI represents an “explosion of opportunity” and could drastically change the landscape of game development. Making a game is already easier than it was: nearly 13,000 titles were published last year on Steam, a games platform, almost double the number in 2017. Gaming may soon resemble the music and video industries in which most new content on Spotify or YouTube is user-generated. One games executive predicts that small firms will be the quickest to work out what new genres are made possible by AI. Last month Raja Koduri, an executive at Intel, left the chip maker to found an AI-gaming startup.

Don’t count the big studios out, though. If they can release half a dozen high-quality titles a year instead of a couple, it might chip away at the hit-driven nature of their business, says Josh Chapman of Konvoy, a gaming focused VC firm. A world of more choices also favors those with big marketing budgets. And the giants may have better answers to the mounting copyright questions around AI. If generative models have to be trained on data to which the developer has the rights, those with big back-catalogues will be better placed than startups. Trent Kaniuga, an artist who has worked on games like “Fortnite”, said last month that several clients had updated their contracts to ban AI-generated art.

If the lawyers don’t intervene, unions might. Studios diplomatically refer to AI assistants as “co-pilots”, not replacements for humans.

7 . The cultivation of plants by ants is more widespread than previously realized, and has evolved on at least 15 separate occasions.

There are more than 200 species of ant in the Americas that farm fungi (真菌) for food, but this trait evolved just once sometime between 45 million and 65 million years ago. Biologists regard the cultivation of fungi by ants as true agriculture appearing earlier than human agriculture because it meets four criteria: the ants plant the fungus, care for it, harvest it and depend on it for food.

By contrast, while thousands of ant species are known to have a wide variety of interdependent relationships with plants, none were regarded as true agriculture. But in 2016, Guillaume Chomicki and Susanne Renner at the University of Munich, Germany, discovered that an ant in Fungi cultivates several plants in a way that meets the four criteria for true agriculture.

The ants collect the seeds of the plants and place them in cracks in the bark of trees. As the plants grow, they form hollow structures called domain that the ants nest in. The ants defecate (排便) at designated absorptive places in these domain, providing nutrients for the plant. In return, as well as shelter, the plant provides food in the form of fruit juice.

This discovery prompted Chomicki and others to review the literature on ant-plant relationships to see if there are other examples of plant cultivation that have been overlooked. “They have never really been looked at in the framework of agriculture,” says Chomicki, who is now at the University of Sheffield in the UK. “It’s definitely widespread.”

The team identified 37 examples of tree-living ants that cultivate plants that grow on trees, known as epiphytes (附生植物). By looking at the family trees of the ant species, the team was able to determine on how many occasions plant cultivation evolved and roughly when. Fifteen is a conservative estimate, says Campbell. All the systems evolved relatively recently, around 1million to 3 million years ago, she says.

Whether the 37 examples of plant cultivation identified by the team count as true agriculture depends on the definitions used. Not all of the species get food from the plants, but they do rely on them for shelter, which is crucial for ants living in trees, says Campbell. So the team thinks the definition of true agriculture should include shelter as well as food.

8 . Investors probably expect that following the suggestions of stock analysts would make them better off than doing the exact opposite. _________, recent research by Nicola Gennaioli and his colleagues shows that the best way to gain excess return s would be to invest in the shares least favored by analysts. They compute that, during the last 35 years, investing in the 10 percent of U. S. stocks analysts were most _________ about would have yielded on average 3 percent a year. _________, investing in the 10 percent of stocks analysts were most pessimistic about would have yielded a surprising 15 percent a year.

Gennaioli and colleagues shed light on this _________ with the help of cognitive sciences and, in particular, using Kahneman and Tversky's concept of representativeness. Decision makers, according to this view, _________ the representative features of a group or a phenomenon. These are defined as the features that occur more frequently in that group than in a baseline reference group.

After observing strong earnings growth—the explanation goes—analysts think that the firm may be the next Google. “Googles” are in fact more frequent among firms experiencing strong growth, which makes them _________. The problem is that “Googles” are very _________ in absolute terms. As a result, expectations become too optimistic, and future performance_________. A model of stock prices in which investor beliefs follow this logic can account both qualitatively and quantitatively for the beliefs of analysts and the dynamics (动态变化) of stock returns.

In related work, the authors also show that the same model can _________ booms and busts in the volume of credit and interest rate spreads.

These works are part of a research project aimed at taking insights from cognitive sciences and at __________them into economic models. Kahneman and Tversky's concept of “representativeness” lies at the heart of this effort. “In a classical example, we __________ to think of Irishmen as redheads because red hair is much more frequent among Irishmen than among the rest of the world,” Prof. Gennaioli says. “However, only 10 percent of Irishmen are redheads. In our work, we develop models of belief formation that show this logic and study the __________ of this important psychological force in different fields.”

Representativeness helps describe __________ and behavior in different fields, not only in financial markets. One such field is the formation of stereotypes about social groups. In a recent experimental paper, Gennaioli and colleagues show that representativeness can explain self-confidence, and in particular the __________ of women to compete in traditionally male subjects, such as mathematics. A slight prevalence of __________ male math ability in the data is enough to make math ability un-representative for women, driving their under confidence in this particular subject.

Japan’s robot revolution in senior care

Japan’s artificial intelligence expertise is transforming the elder care industry, with     1     (specialize) robotic care accomplishing more than just taking pressure off the critical shortage of caregivers. Senior care facilities across Japan are testing out such new robots     2     deliver a collection of social and physical health care and the government-backed initiative has been met with positive reviews by elderly residents.

The rapidly graying population     3     (eye) by the government as a potential market for medical technology now. Disappointing government predictions show that by 2025, Japan's first baby boomers will have turned 75 and about 7 million people are likely to suffer from some form of dementia (痴呆). The nation won't be able to avoid a dementia crisis     4     an additional 380,000 senior care workers.

The long-standing shortage of professional care workers has encouraged the Japanese government     5     (simplify) procedures for foreign caregivers to be trained and certified. The current Technical Intern Training Program between Vietnam, the Philippines, and Indonesia, under     6     Economic Partnership Agreement, was extended to include nursing care as well as agriculture, fishery, and construction sectors.

    7     the government made efforts to increase the numbers of senior care workers, the target number of foreign graduates has still fallen flat, with the national caregiver examination proving a major obstacle to pass. The success rate for foreign students was a merely 106 students last year,     8     has slightly improved to 216 students this year. Another depressing reality is that 19 to 38 percent of foreign nurses who pass the exam opt to leave the industry and return home,    9     (cite) tough work conditions and long hours. Given the challenges, this is     10     the government believes care robots will be able to step in.