評論
打印
Yet what is possible in public health is not always so easy in national security. Western intelligence agencies must contend with laws governing how private data may be gathered and used. In its paper, GCHQ says that it will be mindful of systemic bias, such as whether voice-recognition software is more effective with some groups than others, and transparent about margins of error and uncertainty in its algorithms. American spies say, more vaguely, that they will respect "human dignity, rights, and freedoms". These differences may need to be ironed out. One suggestion made by a recent task-force of former American spooks in a report published by the Centre for Strategic and International Studies (CSIS) in Washington was that the "Five Eyes" intelligence alliance—America, Australia, Britain, Canada and New Zealand—create a shared cloud server on which to store data.
然而，在公共衛(wèi)生領(lǐng)域可行的事情在國家安全領(lǐng)域并不總是那么容易做到。西方情報(bào)機(jī)構(gòu)必須應(yīng)對有關(guān)如何收集和使用私人數(shù)據(jù)的法律。GCHQ在其論文中表示，它會(huì)注意系統(tǒng)偏見，比如語音識(shí)別軟件對某些群體是否比其他群體更有效，以及算法的誤差和不確定性的界限是否透明。美國間諜含糊表示他們將尊重“人的尊嚴(yán)、權(quán)利和自由”。這些分歧可能需要解決。在華盛頓戰(zhàn)略與國際研究中心(CSIS)最近發(fā)布的一份報(bào)告中，前美國特工小組提出了一個(gè)建議：“五眼”情報(bào)聯(lián)盟——美國、澳大利亞、英國、加拿大和新西蘭——可以創(chuàng)建一個(gè)用于存儲(chǔ)數(shù)據(jù)的共享云服務(wù)器。
In any case, the constraints facing AI in intelligence are as much practical as ethical. Machine learning is good at spotting patterns—such as distinctive patterns of mobile-phone use—but poor at predicting individual behaviour. That is especially true when data are scarce, as in counterterrorism. Predictive-policing models can crunch data from thousands of burglaries each year. Terrorist attacks are much rarer, and therefore harder to learn from.
無論如何，人工智能在智能領(lǐng)域面臨的限制既是現(xiàn)實(shí)的，也是道德的。機(jī)器學(xué)習(xí)擅長識(shí)別模式——比如手機(jī)使用的獨(dú)特模式——但在預(yù)測個(gè)人行為方面卻很差。在數(shù)據(jù)匱乏的情況下尤其如此，比如在反恐行動(dòng)中。預(yù)測警務(wù)模型可以處理每年數(shù)千起盜竊案的數(shù)據(jù)。恐怖襲擊要罕見得多，因此也更難從中吸取教訓(xùn)。
That rarity creates another problem, familiar to medics pondering mass-screening programmes for rare diseases. Any predictive model will generate false positives, in which innocent people are flagged for investigation. Careful design can drive the false-positive rate down. But because the "base rate" is lower still—there are, mercifully, very few terrorists—even a well-designed system risks sending large numbers of spies off on wild-goose chases.
這種罕見帶來了另一個(gè)問題，正在考慮對罕見疾病進(jìn)行大規(guī)模篩查的醫(yī)生對這個(gè)問題很熟悉。任何預(yù)測模型都會(huì)出現(xiàn)誤報(bào)，無辜的人被標(biāo)記為調(diào)查對象。精心的設(shè)計(jì)可以降低誤報(bào)率。但是由于“基本比率”仍然較低——幸運(yùn)的是，恐怖分子很少——即使是一個(gè)設(shè)計(jì)良好的系統(tǒng)也有可能使大量間諜在徒勞的追捕中喪命。
And those data that do exist may not be suitable. Data from drone cameras, reconnaissance satellite and intercepted phone calls, for instance, are not currently formatted or labelled in ways that are useful for machine learning. Fixing that is a "tedious, time-consuming, and still primarily human task exacerbated by differing labelling standards across and even within agencies", notes the CSIS report. That may not be quite the sort of work that would-be spies signed up for.
而那些確實(shí)存在的數(shù)據(jù)可能并不合適。例如，來自無人機(jī)攝像頭、偵察衛(wèi)星和截獲的電話的數(shù)據(jù)，目前尚未格式化或標(biāo)記為有利于機(jī)器學(xué)習(xí)的方式。CSIS的報(bào)告指出，解決這一問題是“一項(xiàng)乏味、耗時(shí)且主要還是人工的任務(wù)，各機(jī)構(gòu)甚至各機(jī)構(gòu)內(nèi)部不同的標(biāo)簽化標(biāo)準(zhǔn)加劇了這一問題”。這可能不是那些想成為間諜的人愿意做的工作。
譯文由可可原創(chuàng)，僅供學(xué)習(xí)交流使用，未經(jīng)許可請勿轉(zhuǎn)載。
