onedaylastweekat 9.29 amihunchednervouslyovermykeyboardandpreparedtodobattlewithanentitycalledemma.wewereeachprimedtowriteabit u ATA at 9.30 amandfileourstoriestomyeditor.iwassureemmawouldbequickerthanme是but I really hoped I would be better .最近一天早上9点29分,我开刀我们打算在9点30分写关于英国政府低收入数据的文章,把那篇文章提交给我的编辑。我确认艾玛写得比我快,但我希望她能写得更好。
Her creator,a start-up called Stealth, calls Heran“autonomousartificialintelligence”designedtodeliverprofessionalservicessuchasresearchandanalysis.since it Predicthe arworkersincludingjournalists,I wanted to put it to the test .她的创造者,名为Stealth的初创企业称她为“。现在预计人工智能将代替白领员工(包括记者)流行,所以我想测试一下。emmawasindeedquick : shefiledin 12 minutestomy 35.hercopywasalsobetterthaniexpected.herfactswererightandsheevenincluded antconn FBR EXIT (althoughshewasofthedubiousopinionthatitwouldbea“Tailwind”Fortheukeconomy ).But to My Relief, shelackedthemostimportantjournalisticskillofall : theabilitytodistinguishthenewsworthyfromthedull.whileshecorrectlypoints the nged,sheoverlookedthatthenumberofjobseekershadrisenforthefirsttimeinalmostayear .艾玛显然速度很快:她12分钟内提出。
而且她的文章比我预想的好。她列举的事实是正确的,也包括与英国放弃欧洲(Brexit )等背景相关的可能性(但指出“展开”英国经济的想法各种各样)。
但我深深恳求的是,她缺乏最重要的新闻工作人员的技能。从无聊的资料中提取有新闻价值的内容的能力。
她正确认为失业率不会改变,但无视求职者人数近一年来首次减少。In truth, mostpeoplewhoworkonartificialintelligenceadmitisnotgoingtomakehumansobsoleteanytimesoon.itissimplynotintelligentenoughyes wes en,though,ismoresubtlebutnolessimportant.thelinesarebeginningtoblurbetweenworkdonebyhumansand那还太聪明了。现在再次发生的事情更复杂,但最重要。
人与机器继续执行的工作之间的界限开始变得模糊。For some workers, thiscouldbeaboon.icouldimagineascenariowhereanentitylikeemmacoulddorudimentaryreportsonrepetitivedatarees thensendthemtoahuman fy.theassociatedpressalreadyusesaprogramcalledautomatedinsightstowritesimplecoomple ate results stories.in these cases, humanshavetheadvantage : machinesnotobliteratingthembuttakingovertheboringbitsoftheirjobssotheycanspendmoretimeonthecreative 美联社(Associated Press )已经为名为Automated Insights的计划制作了非常简单的公司业绩报道。在这种情况下,人类具有优势。
机器没有出自人类。让人类把更好的时间花在创新工作和有价值的工作上,代替人类工作的无聊部分。butnotallhumansaremovingupthevaluechain.therearesomeboringtasksatwhichmachinesareverybad.anarmyoflow-paidpeopleare有无聊的工作许多低薪工人默默地跟着这些工作。
taketheworkersonamazon’smechanicalturk是asiterunbytheonlineretailerwhere“requesters”Pay“Turk ers”todosimplemicrotasksthate copyingphotocopiedreceiptsintospreadsheets.Amazoncallsthese“Humanintelligencetasks”,or HITs, andtheytendtopayafewcentsapiece.thenamemechanicalturkcomesfromafakechess-playingmachinefromthe 18 th century 3360 whileitlit ike a a person was secretly hiding inside .以亚马逊土耳其机器人(Mechanical Turk )工人为例,这是该网上零售商运营的网站,“就这样。将与照片相关的关键字作为标签张贴的打印收据加载到电子表格中。亚马逊被称为“人类智能任务”,通常每个任务要花几美分的费用。Mechanical Turk的名字来源于18世纪的欺诈对局机器。
这台机器看起来是自动化的,只是人类躲在机器内部。pinterestisagoodexampleofacompanythatusessiteslikethis.onedeveloperexplainsinarecentbloghowituses“crowd workers”toe es approproprs ches”generatedbyitscomputers.humansarestillbetteratmakingthesejudgmentsthanmaching RTI Ficial’artificial Intelligence,“she conn 一位研究开发人员在最近的博文中说明了如何利用“众包工作者”评价计算机分解的“趋势检索”是否合适。
人类依然比机器更擅长做这些评价。她得出了“创造了‘人类’的人工智能”的结论。someofthenewchatbotandaiservicesalsohavepeoplehidinginside :“humanspretendingtoberobotspretendingtobehumans”asbloombergpu eopleoftenreviewandeditai-generatedresponsesbeforetheyaresent .就像彭博(bloombeent )。
这些人经常在人工智能恢复分解之前展开编辑。What about Emma? Is there a human lurking behind her? Shaunak Khire,co-founder of Stealth,saysemmahasateamofhuman“trainers”butinsiststheoutputisallherown .艾玛呢? 有人躲在她后面吗? Stealth响应创始人Shaunak Khire,艾玛享受着由人类“教练”组成的团队,但他否认她几乎在独立的国家完成了工作。itisalwaysgoingtobehardforlaypeopleusingtheseservicestoknowforsure .用于这些服务的外行很难确认这个。
杰夫bigham,assistantprofessoratcarnegiemellonuniversity’shuman-computerinteractioninstitute是worksasanadvisertoemma’screan for example, bymakingitameanstoacquireskills.heaskshimself :“whatwouldmakemeproudtohavemydaughtergrowuptobeacrowd”even then, itisnotclearhowlongthissortofworkwilllast .卡内基梅隆大学(Carnegie Mellon University )嵌入式研究所(human-computerinters ) 他想想办法让众包不是那么机械,而是有成就感的。例如,将其作为提供技能的方法。他回答说:“有什么能让自己的女儿长大成为众包的骄傲呢?” 尽管如此,我们不确定这样的工作会持续很长时间。
heexplainsthatwhenhumansperformtasksthatmachinescannotyetdo是theycreateanexhaustof“training data”fromwhichaicanlearn.inothes allthosecrowdworkersandchatboteditorsareworkingsteadilytowardstheirownobsolescence .认为,如果人类继续执行机器还不能执行的任务,他们会大量“训练” 换句话说,所有这些众包都是工人和聊天机器人的编辑每天为出局工作。shouldyouwelcomeorfeartheriseofintelligentmachines? That depends on whether they willbe working for you,or you will be working for them .我应该青睐智能机器的崛起还是担心? 这各不相同。他们是为你工作,还是你为他们工作?。
本文来源:开云体育-www.bzttv.com