yabo下载App

亚博yabo下载App

yabo下载App-robotsarenotjusttakingpeople’sjobsaway,they are beginning to hand them out,too .机器人不仅开始偷人类的工作,还开始向人类分发工作。 gotoanyrecruitmentindustryeventandyouwillfindtheairisthickwithtermslikemachinelearning, big data and predictive analytics .参加theargumentforusingthesetoolsinrecruitmentissimple .采用这些工具的理由非常简单。

robo-recruiterscansiftthroughthousandsofjobcandidatesfarmoreefficientlythanhumans .机器人招聘者可慢检查成千上万的应聘者,效率远远高于人类They can also do it more fairly .他们可以变得更公平。 sincetheydonotharbourconsciousorunconscioushumanbiases,theywillrecruitamorediverseandmeritocraticworkforce .像人类一样无意识或有意地持有thisisaseductiveideabutitisalsodangerous .这是一个非常吸引人的主意,但也有危险性。 algorithmsarenotinherentlyneutraljustbecausetheyseetheworldinzerosandones .算法的中立不是唯一的,因为他们看到的世界只不过是0和1。 For a start,anymachinelearningalgorithmisonlyasgoodasthetrainingdatafromwhichitlearns .首先,任何机器学习的算法都比自学的训练数据好。

takethephdthesisofacademicresearchercolinlee,releasedtothepressthisyear.heanalyseddataonthesuccessorfailureof 441, 769 jobapplicationsandbuiltamodelthatcouldpredictwith 70 to 80 percentaccuracywhichcandidateswouldbeinvitedtointerview .学术研究者科林李以今年在媒体上发表的博士论文为例,分析了44.1769万件顺利不顺利的兼职申请人,制作了精度约70%到80%的模型,预测了哪个申请人不会被邀请参加试映。 thepressreleasepluggedthisalgorithmasapotentialtooltoscreenalargenumberofcvswhileavoidinghumanerrorandunconsciousbias .此新闻稿butamodellikethiswouldabsorbanyhumanbiasesatworkintheoriginalrecruitmentdecisions .但是,这样的模型不吸收第一个就业要求中的人为工作场所种族歧视For example,theresearchfoundthatagewasthebiggestpredictorofbeinginvitedtointerview, withtheyoungestandtheoldestapplicantsleastlikelytobesuccessful .例如,在上述研究中,年龄因素可以在程度上预测应聘者是否未被邀请,最大年龄和最长年龄的youmightthinkitfairenoughthatinexperiencedyoungstersdobadly是buttheroutinerejectionofoldercandidatesseemslikesomethingtoinvess 因为没有经验的年轻人打蜡不好。 mrleeacknowledgestheseproblemsandsuggestsitwouldbebettertostripthecvsofattributessuchasgender,age and ethnicity before using them e
In a paper published this year, academicssolonbarocasandrewselbstusetheexampleofanemployerwhowantselectthosecandidatesmostlikelytostayforthelongterm .现在梭伦巴罗斯和安德鲁谢尔博斯的两位学者被用于员工想选择多年来最有可能回到职场的员工的案例。 ifthehistoricaldatashowwomentendtostayinjobsforasignificantlyshortertimethanmen (possiblybecausetheyleavewhentheyhavechildrer ) thealgorithmwillprobablydiscriminateagainstthemonthebasisofattributesthatareareliableproxyforgender .如果历史数据明确, 女性员工在职场停留的时间比男性员工多很多(有了孩子后可能不会辞职),算法有可能利用这些性别指向具体的属性,得到对女性有利的结果。

亚博yabo下载

亚博yabo下载App

orhowaboutthedistanceacandidatelivesfromtheoffice? thatmightwellbeagoodpredictorofattendanceorlongevityatthecompany。 butitcouldalsoinadvertentlydiscriminateagainstsomegroups,sinceneighbourhoodscanhavedifferentethnicorageprofiles .申请人的地址和办公室的但是,由于住宅社区不同,种族和年龄的特征也不同,所以在无意识中也有可能不存在种族歧视的一部分群体。

thesescenariosraisethetrickyquestionofwhetheritiswrongtodiscriminateevenwhenitisrationalandunintended.thisismurkylegalterrres In the US, thedoctrineofdisparateimpactoutlawsostensiblyneutralemploymentpracticesthatdisproportionatelyharmprotectedclasses eveniftheempl iminate .在美国,根据差异影响(disparate impact )的原则,在奇异中立的雇佣实践中损失远远超过受保护阶层的情况下,即使员工不是有意的种族主义也是违法的。 butemployerscansuccessfullydefendthemselvesiftheycanprovethereisastrongbusinesscaseforwhattheyaredoing .但是, 如果员工能证明这种做法有很强的商业理由,iftheintentionofthealgorithmissimplytorecruitthebestpeopleforthejob,that may be a good enough defence Still,itisclearthatemployerswhowantamorediverseworkforcecannotassumethatalltheyneedtodoisturnoverrecruitmenttoacomputer . if TT theywillneedtousedatamoreimaginatively .如果那是他们的想法,他们必须使数据更有想象力。

insteadoftakingtheirowncompanycultureasagivenandlookingforthecandidatesstatisticallymostlikelytoprosperwithinit, for example theycouldseekoutdataaboutwhere (andinwhichcircumstances ) amorediversesetofworkersthrive .例如将他们自己公司的文化调整到另一个既定条件machinelearningwillnotpropelyourworkforceintothefutureiftheonlythingitlearnsfromisyourpast .机器学习唯一能告诉我的只是你的过去,那是你的从政【yabo下载App】。

本文来源:App-www.ahhuidian.com

admin

相关文章