下载掌阅APP，畅读海量书库
立即打开
如果你和家人一起旅行了数千英里，逃离冲突，你会相信计算机程序会找到你的新家吗？
测试中可能遇到的词汇和知识：
pondered['pɒndəd] v.深思
refugee[ˌrefjuˈdʒi:] n.难民
altruism[ˈæltruɪzəm] n.利他主义
阅读马上开始，建议您计算一下阅读整篇文章所用的时间，对照下方的参考值就可以评估出您的英文阅读水平。
------------------------------
If you had travelled thousands of miles with your family, fleeing conflict, would you trust a computer program to find you a new home? I pondered this question after hearing that economists at Oxford University were using artificial intelligence to work out where to settle refugees.
Alarmed by the scale of the Mediterranean refugee crisis, the team — comprising academics in Sweden and the US alongside UK counterparts — designed a program that the US State Department has been using since August. It focuses on finding the best location for a refugee’s job prospects but also takes into account other “needs” such as medical facilities, schools and the proximity of fellow migrants who speak the same language.
Once this has all been weighed up, the program suggests an area match for each migrant. Crucially, the final decision on whether to act on the software’s recommendation of a placement in Tallahassee, Florida, or Paterson, New Jersey, lies with a member of staff in the refugee agency.
So far, the AI mechanism — known affectionately as “Annie” after Annie Moore, the first person to be processed through the US immigration facility at Ellis Island in 1892 — is outperforming humans. Early results show that matches made by Annie increase the refugee employment rate within the first 90 days by about one-third.
Alex Teytelboym, the Oxford University economist who pioneered the program, insists that the intention was “absolutely not” to replace people but to lighten the workload so staff can focus on the trickiest cases. “Eighty per cent of the refugees are very easy to place, they’re straightforward,” he explains. “So there is a lot that can be automated, and it would just be much better if these resettlement staff spent their time on cases that require their attention.”
It is easy to see the advantages of Annie, especially in times of public sector cuts that have hit the UK particularly hard. Staff would have time to consider difficult problems and might also feel less overworked. In the meantime, the more widely the program is used, the more data it collects and the better it becomes. This gives users of the software an incentive to share it as widely as possible, promoting greater altruism between agencies. The Oxford University team, which co-designed Annie, is now looking at other uses of the program, such as matching vulnerable children to foster-care placements.
Not everyone is so keen. Some refugee agency staff were initially worried that an AI program would make their jobs redundant. “People immediately think, ‘You just want to get rid of all of us’,” Teytelboym says. “So the first resistance comes precisely from those you want to help.” They have also been reluctant to use Annie to distribute refugees between the US’s nine different settlement agencies, because this could introduce bias into a system that is currently random. The danger is that some agencies are landed with the “harder-to-settle” refugees, hampering their chances of meeting the State Department’s targets.
More broadly, there will always be concerns around the ethics of entrusting computer programs with decisions that can change the course of someone’s life. AI-based software relies on huge data inputs and, with a system this complex, there can be a lack of transparency about how final conclusions are drawn. I also wonder whether the service users — in this case refugees — have a right to know whether the future pattern of their lives is being dictated by human or machine.
Software designers are not without their own ideological biases. These have the potential to nudge government policy in ways that are either good and bad, depending on your political persuasion. Teytelboym’s vision is that, eventually, Annie’s inputs should include the preferences of refugees themselves, gleaned from a short questionnaire carried out before they arrive in the country.
It is hard to imagine Donald Trump’s administration embracing the idea that refugees’ choices matter. But it’s an exciting prospect that decisions made by a machine could make our future policy more, rather than less, humane.
请根据你所读到的文章内容，完成以下自测题目：
A. Job prospects
B. Medical facilities
C. Schools
D. Communities
答案 (1)
A. It's the name of a previous immigrant
B. It's the inventor's daughter's name
C. It's the name of a US first lady
D. It's unknown to everyone
答案 (2)
A. The aim of the program is to replace all human staff with AI
B. Human involvement is still much necessary
C. The majority of the refugees are very easy to place
D. Placing refugees is hard work, only suitable for AI
答案 (3)
A. Software designers don't have any bias.
B. Since the decision is based on software, it's completely transparent.
C. The refugees' own preference should be included in the AI program.
D. Donald Trump thinks the refugee should decide their own destiny.
答案 (4)
(1) 答案：A解释：It focuses on finding the best location for a refugee’s job prospects but also takes into account other “needs” such as medical facilities, schools and the proximity of fellow migrants who speak the same language.
(2) 答案：A解释：So far, the AI mechanism — known affectionately as “Annie” after Annie Moore, the first person to be processed through the US immigration facility at Ellis Island in 1892 — is outperforming humans.
(3) 答案：C解释：Alex Teytelboym, the Oxford University economist who pioneered the program, insists that the intention was “absolutely not” to replace people but to lighten the workload so staff can focus on the trickiest cases. “Eighty per cent of the refugees are very easy to place, they’re straightforward,” he explains. “So there is a lot that can be automated, and it would just be much better if these resettlement staff spent their time on cases that require their attention.”
(4) 答案：C解释：Teytelboym’s vision is that, eventually, Annie’s inputs should include the preferences of refugees themselves, gleaned from a short questionnaire carried out before they arrive in the country.