{"id":32781,"date":"2019-10-31T21:48:54","date_gmt":"2019-10-31T18:48:54","guid":{"rendered":"https:\/\/prohoster.info\/blog\/o-predvzyatosti-iskusstvennogo-intellekta\/"},"modified":"2019-10-31T21:48:54","modified_gmt":"2019-10-31T18:48:54","slug":"o-predvzyatosti-iskusstvennogo-intellekta","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/o-predvzyatosti-iskusstvennogo-intellekta","title":{"rendered":"Rreth paragjykimit t\u00eb inteligjenc\u00ebs artificiale","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"<p><img decoding=\"async\" alt=\"Rreth paragjykimit t\u00eb inteligjenc\u00ebs artificiale\" src=\"\/wp-content\/uploads\/2019\/04\/05a731677e1231acceff953db8e1e26a.jpg\" style=\"display:block;margin: 0 auto;\" \/><\/p>\n<p><\/p>\n<h2 id=\"tldr\">tl;dr:<\/h2>\n<p><\/p>\n<ul>\n<li>M\u00ebsimi makinerik k\u00ebrkon modele n\u00eb t\u00eb dh\u00ebna. Por inteligjenca artificiale mund t\u00eb jet\u00eb \"e paragjykuar\" \u2014 do t\u00eb thot\u00eb, t\u00eb gjej\u00eb modele t\u00eb gabuara. P\u00ebr shembull, nj\u00eb sistem p\u00ebr zb \ubc1c\uacac eksperimentin e kancerit t\u00eb l\u00ebkur\u00ebs nga fotografia mund t\u00eb jap\u00eb m\u00eb shum\u00eb r\u00ebnd\u00ebsi p\u00ebr imazhet e b\u00ebr\u00eb n\u00eb zyr\u00ebn e mjekut. M\u00ebsimi makinerik nuk di <em>t\u00eb kuptoj\u00eb<\/em>: algoritmet e tij vet\u00ebm zbulojn\u00eb modele n\u00eb numra, dhe n\u00ebse t\u00eb dh\u00ebnat nuk jan\u00eb p\u00ebrfaq\u00ebsuese, rezultati i p\u00ebrpunimit do t\u00eb jet\u00eb i till\u00eb. Dhe kapja e k\u00ebtyre gabimeve mund t\u00eb jet\u00eb e v\u00ebshtir\u00eb p\u00ebr shkak t\u00eb mekanik\u00ebs s\u00eb m\u00ebsimit makinerik. <noindex><a rel=\"nofollow\" name=\"habracut\"><\/a><\/noindex><\/li>\n<li>Fusha m\u00eb e dukshme dhe shqet\u00ebsuese \u00ebsht\u00eb variacioni njer\u00ebzor. Ekzistojn\u00eb shum\u00eb arsye pse t\u00eb dh\u00ebnat mbi njer\u00ebzit mund t\u00eb humbasin objektivitetin edhe n\u00eb faz\u00ebn e mbledhjes. Por nuk duhet t\u00eb mendojm\u00eb se kjo \u00e7\u00ebshtje i p\u00ebrket vet\u00ebm njer\u00ebzve: nj\u00ebsoj v\u00ebshtir\u00ebsi kan\u00eb lindur kur provohet t\u00eb zbulohet nj\u00eb p\u00ebrmbytje n\u00eb nj\u00eb depo apo nj\u00eb turbin\u00eb gazi q\u00eb ka dal\u00eb jasht\u00eb funksionit. Disa sisteme mund t\u00eb ken\u00eb paragjykime p\u00ebr ngjyr\u00ebn e l\u00ebkur\u00ebs, t\u00eb tjer\u00eb do t\u00eb ken\u00eb prirje t\u00eb paragjykuara ndaj sensor\u00ebve Siemens. <\/li>\n<li>K\u00ebto probleme nuk jan\u00eb t\u00eb reja p\u00ebr m\u00ebsimin makinerik, dhe nuk i p\u00ebrkasin vet\u00ebm atij. Supozime t\u00eb gabuara b\u00ebhen n\u00eb \u00e7do struktur\u00eb komplekse, dhe gjithmon\u00eb \u00ebsht\u00eb e v\u00ebshtir\u00eb t\u00eb kuptosh pse \u00ebsht\u00eb marr\u00eb nj\u00eb vendim i caktuar. Duhet t\u00eb luftohet me k\u00ebt\u00eb n\u00eb m\u00ebnyr\u00eb komplekse: t\u00eb krijohen mjete dhe procese p\u00ebr verifikim \u2014 dhe t\u00eb edukohen p\u00ebrdoruesit n\u00eb m\u00ebnyr\u00eb q\u00eb ata t\u00eb mos ndjekin verb\u00ebrisht rekomandimet e IA. M\u00ebsimi makinerik v\u00ebrtet b\u00ebn disa gj\u00ebra shum\u00eb m\u00eb mir\u00eb se ne \u2014 por qent\u00eb, p\u00ebr shembull, jan\u00eb shum\u00eb m\u00eb efektiv\u00eb se njer\u00ebzit n\u00eb zb\u00ebrthimin e narkotik\u00ebve, q\u00eb nuk \u00ebsht\u00eb nj\u00eb arsye p\u00ebr t\u2019i t\u00ebrhequr ata si d\u00ebshmitar\u00eb dhe p\u00ebr t\u00eb nxjerr\u00eb vendime bazuar n\u00eb d\u00ebshmit\u00eb e tyre. Dhe qent\u00eb, p\u00ebr ta th\u00ebn\u00eb, jan\u00eb shum\u00eb m\u00eb t\u00eb zgjuar se \u00e7do sistem m\u00ebsimi makinerik. <\/li>\n<\/ul>\n<p>M\u00ebsimi makinerik sot \u00ebsht\u00eb nj\u00eb nga trendet themelore m\u00eb t\u00eb r\u00ebnd\u00ebsishme teknologjike. \u00cbsht\u00eb nj\u00eb nga m\u00ebnyrat kryesore q\u00eb teknologjia do t\u00eb ndryshoj\u00eb bot\u00ebn p\u00ebr nj\u00eb dekad\u00eb t\u00eb ardhshme. Disa aspekte t\u00eb k\u00ebtyre ndryshimeve shkaktojn\u00eb shqet\u00ebsim. P\u00ebr shembull, ndikimi potencial i m\u00ebsimit makinerik n\u00eb tregun e pun\u00ebs, ose p\u00ebrdorimi i tij p\u00ebr q\u00ebllime anti-etike (p.sh., nga regjimet autoritare). Ka nj\u00eb problem tjet\u00ebr, t\u00eb cilit i kushtohet ky post: <strong>paragjykimi i inteligjenc\u00ebs artificiale<\/strong>. <\/p>\n<p><\/p>\n<p>Kjo \u00ebsht\u00eb nj\u00eb histori e nd\u00ebrlikuar. <\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Rreth paragjykimit t\u00eb inteligjenc\u00ebs artificiale\" src=\"\/wp-content\/uploads\/2019\/04\/d692fbfff47614758ab1d8e1516b2742.jpg\" style=\"display:block;margin: 0 auto;\" \/><br \/>\n<em>AI nga Google mund t\u00eb gjej\u00eb macet. Ky lajm nga viti 2012 ishte di\u00e7ka e ve\u00e7ant\u00eb at\u00ebher\u00eb.<\/em><\/p>\n<p><\/p>\n<h2 id=\"chto-takoe-predvzyatost-ii\">\u00c7far\u00eb \u00ebsht\u00eb \"paragjykimi i AI\"?<\/h2>\n<p><\/p>\n<blockquote><p><em>\"T\u00eb dh\u00ebnat e pap\u00ebrpunuara\" jan\u00eb nj\u00ebkoh\u00ebsisht nj\u00eb oksimoron dhe nj\u00eb ide e keqe; t\u00eb dh\u00ebnat duhet p\u00ebrgatitur mir\u00eb dhe me kujdes.<\/em> \u2014Jeffrey Boker <\/p><\/blockquote>\n<p>Deri diku n\u00eb vitin 2013, p\u00ebr t\u00eb nd\u00ebrtuar nj\u00eb sistem q\u00eb, le t\u00eb themi, njeh macet n\u00eb fotografia, ju duhej t\u00eb p\u00ebrshkruanit hapat logjik\u00eb. Si t\u00eb gjeni qoshet n\u00eb imazh, t\u00eb njoh\u00ebsh syt\u00eb, t\u00eb analizosh teksturat p\u00ebr prezenc\u00ebn e leshit, t\u00eb num\u00ebroni k\u00ebmb\u00ebt, e k\u00ebshtu me radh\u00eb. M\u00eb pas t\u00eb mblidhni t\u00eb gjitha komponent\u00ebt \u2014 dhe t\u00eb zbuloni se asgj\u00eb nga kjo nuk funksionon si\u00e7 duhet. Pothuajse si nj\u00eb kal\u00eb mekanik \u2014 teorikisht mund t\u00eb b\u00ebhet, por n\u00eb praktik\u00eb \u00ebsht\u00eb tep\u00ebr e komplikuar p\u00ebr t'u p\u00ebrshkruar. N\u00eb fund ju keni qindra (ose madje mij\u00ebra) rregulla t\u00eb shkruara me dor\u00eb. Dhe asnj\u00eb model punues. <\/p>\n<p><\/p>\n<p>Me ardhjen e m\u00ebsimit t\u00eb makin\u00ebs, ne ndaluam s\u00eb p\u00ebrdoruri rregullat \"me dor\u00eb\" p\u00ebr njohjen e ndonj\u00eb objekti. N\u00eb vend t\u00eb k\u00ebsaj, ne marrim nj\u00eb mij\u00eb mostrave t\u00eb \"asaj\", X, nj\u00eb mij\u00eb mostrave t\u00eb \"tjet\u00ebr\", Y, dhe e detyrojm\u00eb kompjuterin t\u00eb nd\u00ebrtoj\u00eb nj\u00eb model mbi baz\u00ebn e analiz\u00ebs statistike t\u00eb tyre. M\u00eb pas i japim k\u00ebtij modeli nj\u00eb shembull t\u00eb dh\u00ebnash, dhe ai p\u00ebrcakton me nj\u00eb sakt\u00ebsi t\u00eb caktuar n\u00ebse p\u00ebrputhet me nj\u00ebr\u00ebn nga grupe. <\/p>\n<p><\/p>\n<p>M\u00ebsimi i makin\u00ebs gjeneron nj\u00eb model mbi baz\u00ebn e t\u00eb dh\u00ebnave, dhe jo me an\u00eb t\u00eb nj\u00eb njeriu q\u00eb e shkruan at\u00eb. Rezultatet jan\u00eb impresionuese, ve\u00e7an\u00ebrisht n\u00eb fush\u00ebn e njohjes s\u00eb imazheve dhe modeleve, dhe pik\u00ebrisht p\u00ebr k\u00ebt\u00eb arsye, e gjith\u00eb industria teknologjike po kalon n\u00eb m\u00ebsimin e makin\u00ebs (ML).<\/p>\n<p><\/p>\n<p>Por nuk \u00ebsht\u00eb kaq e thjesht\u00eb. N\u00eb bot\u00ebn reale, mij\u00ebra shembuj tuaj X ose Y gjithashtu p\u00ebrmbajn\u00eb A, B, J, L, O, R dhe madje L. Ata mund t\u00eb jen\u00eb t\u00eb shp\u00ebrndar\u00eb n\u00eb m\u00ebnyr\u00eb t\u00eb pabarabart\u00eb, dhe disa nga ata mund t\u00eb ndodhen aq shpesh, saq\u00eb sistemi do t'i kushtoj\u00eb m\u00eb shum\u00eb v\u00ebmendje atyre sesa objekteve q\u00eb ju interesoni. <noindex><a rel=\"nofollow\" href=\"http:\/\/aiweirdness.com\/post\/171451900302\/do-neural-nets-dream-of-electric-sheep\">\u00c7far\u00eb do t\u00eb thot\u00eb kjo n\u00eb praktik\u00eb? Shembulli im i preferuar \u00ebsht\u00eb kur sistemet e njohjes s\u00eb imazheve<\/a><\/noindex>shikojn\u00eb nj\u00eb kod\u00ebr me bar dhe thon\u00eb: \"dele\". <\/p>\n<p><\/p>\n<p>. E kuptojm\u00eb pse: shumica e fotografive-most\u00ebr \"dele\" jan\u00eb marr\u00eb n\u00eb livadhe, ku ato jetojn\u00eb, dhe n\u00eb k\u00ebto imazhe, bar \u00ebsht\u00eb n\u00eb m\u00ebnyr\u00eb t\u00eb dukshme z\u00ebvend\u00ebsuesi m\u00eb i madh, sesa t\u00eb voglat t\u00eb bardha dhe me lesh, dhe pik\u00ebrisht barin sistemet e konsiderojn\u00eb si m\u00eb t\u00eb r\u00ebnd\u00ebsishmin. <noindex><a rel=\"nofollow\" href=\"https:\/\/www.medpagetoday.com\/dermatology\/generaldermatology\/70513\">projekt<\/a><\/noindex> p\u00ebr identifikimin e kancerit t\u00eb l\u00ebkur\u00ebs n\u00eb fotografi. Doli se dermatolog\u00ebt shpesh fotografonin nj\u00eb linj\u00eb s\u00eb bashku me shenjat e kancerit t\u00eb l\u00ebkur\u00ebs, p\u00ebr t\u00eb regjistruar madh\u00ebsin\u00eb e formacioneve. N\u00eb shembujt e fotografive t\u00eb l\u00ebkur\u00ebs s\u00eb sh\u00ebndosh\u00eb nuk ka linjash. P\u00ebr sistemin e AI, k\u00ebto linja (n\u00eb t\u00eb v\u00ebrtet\u00eb, pixel-et q\u00eb ne i p\u00ebrcaktojm\u00eb si 'linj\u00eb') jan\u00eb b\u00ebr\u00eb nj\u00eb nga dallimet mes grupeve t\u00eb shembujve, ndonj\u00ebher\u00eb m\u00eb t\u00eb r\u00ebnd\u00ebsishme se disa shp\u00ebrthime t\u00eb vogla n\u00eb l\u00ebkur\u00eb. K\u00ebshtu, sistemi i krijuar p\u00ebr t\u00eb njohur kancerin e l\u00ebkur\u00ebs, ndonj\u00ebher\u00eb e njihej linja n\u00eb vend t\u00eb tij. <\/p>\n<p><\/p>\n<p>Pika ky\u00e7e k\u00ebtu \u00ebsht\u00eb se sistemi nuk ka kuptim semantik t\u00eb asaj q\u00eb po shikon. Ne shohim nj\u00eb grup pixel-esh dhe shohim n\u00eb to nj\u00eb dele, l\u00ebkur\u00eb ose linj\u00eb, nd\u00ebrsa sistemi sheh vet\u00ebm nj\u00eb varg numrash. Ai nuk sheh hap\u00ebsir\u00ebn tre-dimensionale, nuk sheh as objekte, as textura, as dele. Ai thjesht sheh modele n\u00eb t\u00eb dh\u00ebna. <\/p>\n<p><\/p>\n<p>Sfidat e diagnoz\u00ebs s\u00eb k\u00ebtyre problemeve jan\u00eb se rrjeti nervor (modeli i gjeneruar nga sistemi juaj i m\u00ebsimit t\u00eb makineris\u00eb) p\u00ebrb\u00ebhet nga mij\u00ebra qindra mij\u00ebra nyjesh. Nuk ka nj\u00eb m\u00ebnyr\u00eb t\u00eb thjesht\u00eb p\u00ebr t\u00eb hedhur nj\u00eb sy n\u00eb model dhe p\u00ebr t\u00eb par\u00eb si merr vendime. Prania e nj\u00eb m\u00ebnyre t\u00eb till\u00eb do t\u00eb n\u00ebnkuptonte se procesi \u00ebsht\u00eb mjaft i thjesht\u00eb p\u00ebr t\u00eb p\u00ebrshkruar t\u00eb gjitha rregullat manualisht, pa p\u00ebrdorur m\u00ebsimin e makineris\u00eb. Njer\u00ebzit shqet\u00ebsohen se m\u00ebsimi i makineris\u00eb ka b\u00ebr\u00eb nj\u00eb lloj 'kutie t\u00eb zez\u00eb'. (Do ta shpjegoj pak m\u00eb von\u00eb pse kjo krahasim \u00ebsht\u00eb paksa ekstrem.) <\/p>\n<p><\/p>\n<p>Kjo, n\u00eb terma t\u00eb p\u00ebrgjithsh\u00ebm, \u00ebsht\u00eb probleme e paragjykimit t\u00eb inteligjenc\u00ebs artificiale ose m\u00ebsimit t\u00eb makineris\u00eb: sistemi p\u00ebr gjetjen e modeleve n\u00eb t\u00eb dh\u00ebna mund t\u00eb gjej\u00eb modele t\u00eb gabuara, dhe ju mund t\u00eb mos e vini re at\u00eb. Kjo \u00ebsht\u00eb nj\u00eb karakteristik\u00eb themelore e teknologjis\u00eb, dhe \u00ebsht\u00eb e dukshme p\u00ebr t\u00eb gjith\u00eb ata q\u00eb punojn\u00eb me t\u00eb n\u00eb qarqet shkencore dhe n\u00eb kompanit\u00eb e m\u00ebdha teknologjike. Por pasojat e saj jan\u00eb komplekse, dhe zgjidhjet tona t\u00eb mundshme p\u00ebr k\u00ebto pasoja gjithashtu. <\/p>\n<p><\/p>\n<p>Le t\u00eb flasim fillimisht p\u00ebr pasojat. <\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Rreth paragjykimit t\u00eb inteligjenc\u00ebs artificiale\" src=\"\/wp-content\/uploads\/2019\/04\/43748412bd82ae53a9c2e47afec9c655.jpg\" style=\"display:block;margin: 0 auto;\" \/><br \/>\n<em>AI mund t\u00eb b\u00ebj\u00eb zgjedhje p\u00ebr ne n\u00eb m\u00ebnyr\u00eb t\u00eb pap\u00ebrshkruar p\u00ebr kategorit\u00eb e caktuara t\u00eb njer\u00ebzve, duke u bazuar n\u00eb nj\u00eb sasi t\u00eb madhe sinjalesh t\u00eb padukshme.<\/em><\/p>\n<p><\/p>\n<h2 id=\"scenarii-predvzyatosti-ii\">Skenar\u00ebt e paragjykimit t\u00eb AI<\/h2>\n<p><\/p>\n<p>E dukshme dhe shqet\u00ebsuese \u00ebsht\u00eb se kjo problem mund t\u00eb shfaqet kur b\u00ebhet fjal\u00eb p\u00ebr larmishm\u00ebrin\u00eb njer\u00ebzore. S\u00eb fundmi, <noindex><a rel=\"nofollow\" href=\"https:\/\/www.reuters.com\/article\/us-amazon-com-jobs-automation-insight\/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G\">ka pasur thashetheme<\/a><\/noindex>, q\u00eb Amazon p\u00ebrpiqej t\u00eb ndihmonte nj\u00eb sistem t\u00eb m\u00ebsimit t\u00eb makinerive p\u00ebr skanimin e par\u00eb t\u00eb kandidat\u00ebve p\u00ebr pun\u00eb. Duke qen\u00eb se midis punonj\u00ebsve t\u00eb Amazonit ka m\u00eb shum\u00eb burra, shembujt e \u201epun\u00ebsimeve t\u00eb suksesshme\u201c gjithashtu ishin kryesisht t\u00eb seksit mashkull, dhe n\u00eb p\u00ebrzgjedhjen e CV-ve, sistemit i ofruan m\u00eb shum\u00eb burra. Amazon e vuri re k\u00ebt\u00eb dhe nuk e l\u00ebshoi sistemin n\u00eb prodhim. <\/p>\n<p><\/p>\n<p>Gj thingesht m\u00eb e r\u00ebnd\u00ebsishme n\u00eb k\u00ebt\u00eb shembull \u00ebsht\u00eb se sistemi, sipas thashethemeve, preferonte kandidat\u00ebt e seksit mashkull, pavar\u00ebsisht se kjo nuk ishte e specificuar n\u00eb CV.<strong> Sistemi shihte modele t\u00eb tjera n\u00eb shembujt e \u201epun\u00ebsimit t\u00eb suksessh\u00ebm\u201c: p\u00ebr shembull, grat\u00eb mund t\u00eb p\u00ebrdorin fjal\u00eb t\u00eb ve\u00e7anta p\u00ebr t\u00eb p\u00ebrshkruar arritjet, ose t\u00eb ken\u00eb hobin e ve\u00e7ant\u00eb. Natyrisht, sistemi nuk e dinte se \u00e7far\u00eb ishte \u201ehokej\u201c, as kush ishin \u201enjer\u00ebzit\u201c, as \u00e7far\u00eb ishte \u201esuksesi\u201c; thjesht b\u00ebnte nj\u00eb analiz\u00eb statistikore t\u00eb tekstit.<\/strong> Por ligj\u00ebsit\u00eb q\u00eb ajo sheh, do t\u00eb mbeteshin ndoshta t\u00eb paq\u00ebndrueshme nga nj\u00eb njeri, dhe disa prej tyre (p\u00ebr shembull, se si njer\u00ebzit e seksit t\u00eb ndrysh\u00ebm e p\u00ebrshkruajn\u00eb suksesin ndryshe) ne, me siguri, do t\u00eb na kishte v\u00ebshtir\u00eb t'i shihnim, edhe duke i par\u00eb ato. <\/p>\n<p><\/p>\n<p>M\u00eb tej, situata b\u00ebhet m\u00eb e komplikuar. Nj\u00eb sistem i m\u00ebsimit t\u00eb makinerive, i cili gjen shum\u00eb mir\u00eb kancerin n\u00eb l\u00ebkur\u00ebn e bardh\u00eb, mund t\u00eb ket\u00eb performanc\u00eb m\u00eb t\u00eb dob\u00ebt me l\u00ebkur\u00ebn e err\u00ebt, ose anasjelltas. Nuk \u00ebsht\u00eb domosdoshm\u00ebrisht p\u00ebr shkak t\u00eb paragjykimit, por sepse, me siguri, duhet t\u00eb nd\u00ebrtohet nj\u00eb model i ve\u00e7ant\u00eb p\u00ebr ngjyr\u00ebn tjet\u00ebr t\u00eb l\u00ebkur\u00ebs, duke zgjedhur karakteristika t\u00eb tjera.<strong> Sistemet e m\u00ebsimit t\u00eb makinerive nuk jan\u00eb t\u00eb nd\u00ebrshtueshme as n\u00eb nj\u00eb fush\u00eb kaq t\u00eb ngusht\u00eb si njohja e imazheve.<\/strong> Duhet t\u00eb konfiguroni sistemin, ndonj\u00ebher\u00eb thjesht duke provuar dhe gabuar, p\u00ebr t\u00eb p\u00ebrmir\u00ebsuar sh\u00ebnimin e ve\u00e7orive n\u00eb t\u00eb dh\u00ebnat q\u00eb ju interesojn\u00eb, derisa t\u00eb arrini sakt\u00ebsin\u00eb e d\u00ebshiruar. Por mund t\u00eb mos e vini re se sistemi \u00ebsht\u00eb 98% i sakt\u00eb n\u00eb pun\u00ebn me nj\u00eb grup dhe vet\u00ebm 91% (pavar\u00ebsisht se kjo \u00ebsht\u00eb m\u00eb e sakt\u00eb se analiza e kryer nga njeriu) me nj\u00eb tjet\u00ebr. <\/p>\n<p><\/p>\n<p>Derim\u00eb tani kam p\u00ebrdorur kryesisht shembuj q\u00eb lidhen me njer\u00ebzit dhe karakteristikat e tyre. Kjo tem\u00eb \u00ebsht\u00eb thelb\u00ebsore p\u00ebr diskutimin rreth k\u00ebsaj \u00e7\u00ebshtjeje. Por \u00ebsht\u00eb e r\u00ebnd\u00ebsishme t\u00eb kuptohet se paragjykimi ndaj njer\u00ebzve \u00ebsht\u00eb vet\u00ebm nj\u00eb pjes\u00eb e problematik\u00ebs. Ne do t\u00eb p\u00ebrdorim m\u00ebsimin e makinerive p\u00ebr shum\u00eb gj\u00ebra, dhe gabimi i mostrave do t\u00eb jet\u00eb i r\u00ebnd\u00ebsish\u00ebm p\u00ebr t\u00eb gjitha ato. Nga ana tjet\u00ebr, n\u00ebse punoni me njer\u00ebz, paragjykimi i t\u00eb dh\u00ebnave mund t\u00eb mos lidhet me ta. <\/p>\n<p><\/p>\n<p>P\u00ebr t\u00eb kuptuar k\u00ebt\u00eb, le t\u00eb kthehemi te shembulli i kancerit t\u00eb l\u00ebkur\u00ebs dhe t\u00eb shqyrtojm\u00eb tri mund\u00ebsi hipotezike p\u00ebr thyerjen e sistemit. <\/p>\n<p><\/p>\n<ol>\n<li>Distribucioni i paansh\u00ebm i njer\u00ebzve: nj\u00eb num\u00ebr i pavendosur fotografish t\u00eb l\u00ebkur\u00ebs me nuanca t\u00eb ndryshme, \u00e7ka \u00e7on n\u00eb rezultate false pozitive ose negative t\u00eb lidhura me pigmentimin. <\/li>\n<li>T\u00eb dh\u00ebnat me t\u00eb cilat trajnohet sistemi p\u00ebrmbajn\u00eb nj\u00eb karakteristik\u00eb t\u00eb zakonshme dhe t\u00eb paanshme q\u00eb nuk \u00ebsht\u00eb e lidhur me njer\u00ebzit dhe nuk ka vler\u00eb diagnostike: nj\u00eb ruler n\u00eb fotografit\u00eb e shfaqjeve t\u00eb kancerit t\u00eb l\u00ebkur\u00ebs ose barin n\u00eb fotografit\u00eb e deleve. N\u00eb k\u00ebt\u00eb rast, rezultati do t\u00eb ndryshoj\u00eb n\u00ebse sistemi n\u00eb imazh do t\u00eb gjej\u00eb piksel q\u00eb njeriu do ta p\u00ebrcaktonte si 'ruler'. <\/li>\n<li>T\u00eb dh\u00ebnat p\u00ebrmbajn\u00eb nj\u00eb karakteristik\u00eb t\u00eb huaj, q\u00eb njeriut nuk i shfaqet, madje edhe n\u00ebse e k\u00ebrkon. <\/li>\n<\/ol>\n<p><\/p>\n<p>\u00c7far\u00eb do t\u00eb thot\u00eb kjo? Ne e dim\u00eb paraprakisht se t\u00eb dh\u00ebnat mund t\u00eb p\u00ebrfaq\u00ebsojn\u00eb n\u00eb m\u00ebnyra t\u00eb ndryshme grupe t\u00eb ndryshme njer\u00ebzish, dhe minimalisht mund t\u00eb planifikojm\u00eb k\u00ebrkimin e p\u00ebrjashtimeve t\u00eb tilla. Me fjal\u00eb t\u00eb tjera, ka shum\u00eb arsye sociale p\u00ebr t\u00eb supozuar se t\u00eb dh\u00ebnat p\u00ebr grupet e njer\u00ebzve tashm\u00eb p\u00ebrmbajn\u00eb nj\u00ebfar\u00eb paragjykimi. N\u00ebse e shohim fotografin\u00eb me ruler, do ta shohim at\u00eb ruler \u2014 thjesht e kemi injoruar m\u00eb par\u00eb, duke e ditur se nuk ka r\u00ebnd\u00ebsi, dhe duke harruar se sistemi nuk di asgj\u00eb.<\/p>\n<p><\/p>\n<p>Por \u00e7far\u00eb ndodh n\u00ebse t\u00eb gjitha fotografit\u00eb tuaja t\u00eb l\u00ebkur\u00ebs s\u00eb pap\u00ebrshtatshme jan\u00eb b\u00ebr\u00eb n\u00eb nj\u00eb zyr\u00eb ku p\u00ebrdoren llambat incandescente, dhe ato t\u00eb l\u00ebkur\u00ebs s\u00eb sh\u00ebndetshme jan\u00eb b\u00ebr\u00eb n\u00ebn drit\u00ebn fluorescente? \u00c7far\u00eb ndodh n\u00ebse, pasi keni p\u00ebrfunduar xhirimin e l\u00ebkur\u00ebs s\u00eb sh\u00ebndetshme, keni p\u00ebrdit\u00ebsuar sistemin operativ n\u00eb telefonin tuaj dhe Apple ose Google kan\u00eb ndryshuar pak algoritmin e zbutjes s\u00eb zhurm\u00ebs? Njeriut nuk mund t'i bie n\u00eb sy kjo, sa her\u00eb ta k\u00ebrkoj\u00eb. Por sistemi i p\u00ebrdorimit t\u00eb makinave do ta shoh\u00eb dhe do ta p\u00ebrdor\u00eb menj\u00ebher\u00eb. Ai nuk di asgj\u00eb. <\/p>\n<p><\/p>\n<p>Nd\u00ebrsa flisnim p\u00ebr lidhje t\u00eb gabuara, ndodhin raste kur t\u00eb dh\u00ebnat jan\u00eb t\u00eb sakta dhe rezultatet jan\u00eb t\u00eb drejta, por ju nuk d\u00ebshironi t\u2019i p\u00ebrdorni ato p\u00ebr arsye etike, ligjore ose menaxheriale. N\u00eb disa juridiksione, p\u00ebr shembull, nuk lejohet t\u00eb ofrohet zbritje p\u00ebr grat\u00eb n\u00eb sigurim, megjith\u00ebse grat\u00eb ndoshta drejtojn\u00eb m\u00eb t\u00eb sigurt. Ne mund ta imagjinojm\u00eb leht\u00ebsisht nj\u00eb sistem q\u00eb, kur analizon t\u00eb dh\u00ebnat historike, i jep grave nj\u00eb koeficient risku m\u00eb t\u00eb ul\u00ebt. Mir\u00eb, le t\u2019i heqim emrat nga mostra. Por kujtoni shembullin e Amazonit: sistemi mund t\u00eb identifikoj\u00eb gjinin\u00eb sipas faktor\u00ebve t\u00eb tjer\u00eb (edhe pse ai nuk e di \u00e7far\u00eb \u00ebsht\u00eb gjinia, as \u00e7far\u00eb \u00ebsht\u00eb makina), dhe ju nuk do ta v\u00ebreni derisa rregullatori t\u00eb analizoj\u00eb pas faktit tarifat q\u00eb i keni propozuar dhe t\u00eb marr\u00eb gjob\u00eb nga ju. <\/p>\n<p><\/p>\n<p>N\u00eb fund, shpesh n\u00ebnkuptohet se do t'i p\u00ebrdorim k\u00ebto sisteme vet\u00ebm p\u00ebr projekte q\u00eb lidhen me njer\u00ebzit dhe nd\u00ebrveprimet sociale. Kjo nuk \u00ebsht\u00eb e v\u00ebrtet\u00eb. N\u00ebse po prodhoni turbinat me gaz, me siguri do t\u00eb d\u00ebshironit t\u00eb aplikoni m\u00ebsimin e makineris\u00eb p\u00ebr telemetrin\u00eb e d\u00ebrguar nga dhjet\u00ebra ose qindra sensor\u00eb n\u00eb produktin tuaj (sensor\u00eb audio, video, temperatur\u00eb dhe \u00e7do sensor tjet\u00ebr gjeneron t\u00eb dh\u00ebna q\u00eb mund t\u00eb p\u00ebrshtaten leht\u00ebsisht p\u00ebr t\u00eb krijuar nj\u00eb model m\u00ebsimi makinerie). Hipotetikisht, mund t\u00eb thoni: \"Ja t\u00eb dh\u00ebnat p\u00ebr nj\u00eb mij\u00eb turbina q\u00eb kan\u00eb d\u00ebshtuar, t\u00eb regjistruara para d\u00ebshtimit t\u00eb tyre, dhe ja t\u00eb dh\u00ebnat nga nj\u00eb mij\u00eb turbina q\u00eb nuk kan\u00eb d\u00ebshtuar. Nd\u00ebrtoni nj\u00eb model p\u00ebr t\u00eb th\u00ebn\u00eb se \u00e7far\u00eb ndryshimi ka mes tyre\". Tani imagjinoni q\u00eb sensor\u00ebt Siemens jan\u00eb n\u00eb 75% t\u00eb turbina t\u00eb k\u00ebqija dhe vet\u00ebm n\u00eb 12% t\u00eb turbina t\u00eb mira (nuk ka lidhje me d\u00ebshtimet). Sistemi do t\u00eb nd\u00ebrtoj\u00eb nj\u00eb model p\u00ebr t\u00eb gjetur turbinat me sensor\u00ebt Siemens. Ups!<\/p>\n<p><\/p>\n<p><img decoding=\"async\" alt=\"Rreth paragjykimit t\u00eb inteligjenc\u00ebs artificiale\" src=\"\/wp-content\/uploads\/2019\/04\/e8097e4913756fdf04826a6628bc6836.jpg\" style=\"display:block;margin: 0 auto;\" \/><br \/>\nImazhi \u2014 Moritz Hardt, UC Berkeley<\/p>\n<p><\/p>\n<h2 id=\"upravlenie-predvzyatostyu-ii\">Menaxhimi i paragjykimeve t\u00eb AI<\/h2>\n<p><\/p>\n<p>\u00c7far\u00eb mund t\u00eb b\u00ebjm\u00eb me k\u00ebt\u00eb? Mund t\u00eb qasemi n\u00eb k\u00ebt\u00eb \u00e7\u00ebshtje nga tri an\u00eb:<\/p>\n<p><\/p>\n<ol>\n<li>Rigorozitet metodologjik n\u00eb mbledhjen dhe menaxhimin e t\u00eb dh\u00ebnave p\u00ebr trajnimin e sistemit.<\/li>\n<li>Veglat teknike p\u00ebr analiz\u00ebn dhe diagnostikimin e sjelljes s\u00eb modelit.<\/li>\n<li>Trajnimi, arsimimi dhe kujdesi gjat\u00eb integrimit t\u00eb m\u00ebsimit t\u00eb makineris\u00eb n\u00eb produkte.<\/li>\n<\/ol>\n<p><\/p>\n<p>N\u00eb librin \"Burgimtar n\u00eb aristokraci\" t\u00eb Molierit ka nj\u00eb shaka: nj\u00eb burr\u00eb u tha se literatura ndahet n\u00eb proz\u00eb dhe poezi, dhe ai me habi zbulon se gjat\u00eb gjith\u00eb jet\u00ebs ka folur proz\u00eb, pa e ditur. Ndoshta, statistik\u00ebt ndihen k\u00ebshtu sot: pa e kuptuar at\u00eb, ata ia kan\u00eb kushtuar karrierat inteligjenc\u00ebs artificiale dhe gabimit t\u00eb mostrave. T\u00eb k\u00ebrkosh gabimin e mostrave dhe t\u00eb shqet\u00ebsohesh p\u00ebr t\u00eb \u2014 kjo nuk \u00ebsht\u00eb nj\u00eb problem i ri, thjesht na nevojitet nj\u00eb qasje sistematike p\u00ebr ta zgjidhur. Si\u00e7 u p\u00ebrmend m\u00eb par\u00eb, n\u00eb disa raste \u00ebsht\u00eb v\u00ebrtet m\u00eb e leht\u00eb ta b\u00ebsh k\u00ebt\u00eb duke studiuar problemet q\u00eb lidhen me t\u00eb dh\u00ebnat p\u00ebr njer\u00ebzit. Ne presim q\u00eb mund t\u00eb kemi paragjykime ndaj grupeve t\u00eb ndryshme t\u00eb njer\u00ebzve, por paragjykimi ndaj sensor\u00ebve Siemens \u00ebsht\u00eb di\u00e7ka q\u00eb na v\u00ebshtir\u00ebson ta imagjinojm\u00eb.<\/p>\n<p><\/p>\n<p>E reja n\u00eb t\u00eb gjitha k\u00ebto \u00ebsht\u00eb, sigurisht, se njer\u00ebzit nuk merren m\u00eb drejtp\u00ebrdrejt me analiz\u00ebn statistikore. Ajo b\u00ebhet nga makinat, t\u00eb cilat krijojn\u00eb modele t\u00eb m\u00ebdha komplekse, t\u00eb nd\u00ebrlikuara p\u00ebr t'u kuptuar. Problemi i transparenc\u00ebs \u00ebsht\u00eb nj\u00eb nga aspektet kryesore t\u00eb problemeve me paragjykimin. Na frik\u00ebson q\u00eb sistemi jo vet\u00ebm q\u00eb \u00ebsht\u00eb i paragjykuar, por nuk ka asnj\u00eb mund\u00ebsi p\u00ebr ta zbuluar at\u00eb paragjykim, dhe se m\u00ebsimi statutor dallohet p\u00ebr k\u00ebt\u00eb nga format e tjera t\u00eb automatizimit, t\u00eb cilat supozohet se p\u00ebrb\u00ebhen nga hapa logjik\u00eb t\u00eb qart\u00eb, q\u00eb mund t\u00eb kontrollohen. <\/p>\n<p><\/p>\n<p>K\u00ebtu ka dy probleme. Ndoshta ne mund t\u00eb kryejm\u00eb nj\u00eb auditim t\u00eb caktuar t\u00eb sistemeve t\u00eb m\u00ebsimit makinerik. Dhe auditimi i \u00e7do sistemi tjet\u00ebr n\u00eb fakt nuk \u00ebsht\u00eb aspak m\u00eb i leht\u00eb.<\/p>\n<p><\/p>\n<p>S\u00eb pari, nj\u00eb nga drejtp\u00ebrdrejtimet e k\u00ebrkimeve moderne n\u00eb fush\u00ebn e m\u00ebsimit makinerik \u00ebsht\u00eb k\u00ebrkimi i metodave p\u00ebr t\u00eb identifikuar funksionalitetin e r\u00ebnd\u00ebsish\u00ebm t\u00eb sistemeve t\u00eb m\u00ebsimit makinerik. N\u00eb k\u00ebt\u00eb rast, m\u00ebsimi makinerik (n\u00eb gjendjen e tij aktuale) \u00ebsht\u00eb nj\u00eb fush\u00eb e re shkencore, q\u00eb po ndryshon shpejt, k\u00ebshtu q\u00eb nuk duhet menduar se gj\u00ebrat e pamundura sot nuk mund t\u00eb b\u00ebhen shum\u00eb reale n\u00eb t\u00eb ardhmen. Projekti <noindex><a rel=\"nofollow\" href=\"https:\/\/openai.com\/\">OpenAI<\/a><\/noindex> \u2014 \u00ebsht\u00eb nj\u00eb shembull interesant p\u00ebr k\u00ebt\u00eb. <\/p>\n<p><\/p>\n<p>S\u00eb dyti, ideja se mund t\u00eb kontrollosh dhe kuptosh procesin e vendimmarrjes n\u00eb sistemet ose organizatat ekzistuese \u00ebsht\u00eb e mir\u00eb n\u00eb teori, por \u00ebsht\u00eb e v\u00ebshtir\u00eb n\u00eb praktik\u00eb. T\u00eb kuptosh se si merren vendimet n\u00eb nj\u00eb organizat\u00eb t\u00eb madhe nuk \u00ebsht\u00eb aspak e leht\u00eb. Madje, edhe n\u00ebse ekziston nj\u00eb proces formal vendimmarrjeje, ai nuk reflekton se si njer\u00ebzit nd\u00ebrveprojn\u00eb n\u00eb t\u00eb v\u00ebrtet\u00eb, dhe ata vet\u00eb shpesh nuk kan\u00eb nj\u00eb qasje sistematike logjike p\u00ebr marrjen e vendimeve t\u00eb tyre. Si\u00e7 tha kolegu im <noindex><a rel=\"nofollow\" href=\"https:\/\/en.wikipedia.org\/wiki\/Vijay_S._Pande\">Vijay Panda<\/a><\/noindex>, <strong>njer\u00ebzit jan\u00eb gjithashtu kuti t\u00eb zeza<\/strong>. <\/p>\n<p><\/p>\n<p>Merrni nj\u00eb mij\u00eb njer\u00ebz n\u00eb disa kompani dhe institute q\u00eb prek\u00ebn nj\u00ebri-tjetrin, dhe problemi b\u00ebhet akoma m\u00eb i komplikuar. Ne e dim\u00eb pas faktit se \"Space Shuttle\" ishte i destinuar t\u00eb shp\u00ebrb\u00ebhej n\u00eb kthim, dhe individ\u00eb t\u00eb ve\u00e7ant\u00eb brenda NASA kishin informacion q\u00eb i jepte atyre arsyen t\u00eb mendonin se mund t\u00eb ndodhte di\u00e7ka e keqe, por sistemi <em>si nj\u00eb e t\u00ebr\u00eb<\/em> nuk e dinte k\u00ebt\u00eb. NASA sapo kishte kaluar nj\u00eb auditim t\u00eb ngjash\u00ebm, duke humbur shat\u00ebn e m\u00ebparshme, dhe megjithat\u00eb ajo humbi edhe nj\u00eb tjet\u00ebr \u2014 p\u00ebr nj\u00eb arsye t\u00eb ngjashme. \u00cbsht\u00eb e leht\u00eb t\u00eb thuash se organizatat dhe njer\u00ebzit ndjekin rregulla logjike t\u00eb qarta, t\u00eb cilat mund t\u00eb kontrollohen, kuptohen dhe ndryshohen \u2014 por p\u00ebrvoja e tregon t\u00eb kund\u00ebrt\u00ebn. Kjo \u00ebsht\u00eb \"<noindex><a rel=\"nofollow\" href=\"https:\/\/ru.wikipedia.org\/wiki\/%D0%93%D0%BE%D1%81%D0%BF%D0%BB%D0%B0%D0%BD_%D0%A1%D0%A1%D0%A1%D0%A0\">iluzionin e Gosplanit<\/a><\/noindex>\u00bb.<\/p>\n<p><\/p>\n<p><strong>Shpesh e krahasoj m\u00ebsimin e makinerive me bazat e t\u00eb dh\u00ebnave, sidomos me ato rrelacionale \u2014 nj\u00eb teknologji themelore e re q\u00eb ndryshoi mund\u00ebsit\u00eb e informatik\u00ebs dhe bot\u00ebs p\u00ebrreth saj, e cila b\u00ebri pjes\u00eb t\u00eb gjith\u00e7kaje, dhe ne e p\u00ebrdorim vazhdimisht pa e kuptuar k\u00ebt\u00eb.<\/strong> N\u00eb bazat e t\u00eb dh\u00ebnave ka gjithashtu probleme, dhe ato kan\u00eb natyr\u00eb t\u00eb ngjashme: sistemi mund t\u00eb nd\u00ebrtohet mbi supozita t\u00eb gabuara ose mbi t\u00eb dh\u00ebna t\u00eb k\u00ebqija, por kjo do t\u00eb jet\u00eb e v\u00ebshtir\u00eb p\u00ebr t'u v\u00ebrejtur, dhe njer\u00ebzit q\u00eb p\u00ebrdorin sistemin do t\u00eb b\u00ebjn\u00eb at\u00eb q\u00eb ai u thot\u00eb, pa b\u00ebr\u00eb pyetje. Ekziston nj\u00eb mori shakash t\u00eb vjetra rreth punonj\u00ebsve t\u00eb taksave, t\u00eb cil\u00ebt dikur e regjistruan gabimisht emrin tuaj, dhe t\u00eb bind\u00ebsh ata t\u00eb korrektojn\u00eb gabimin \u00ebsht\u00eb shum\u00eb m\u00eb e v\u00ebshtir\u00eb se sa t\u00eb ndryshosh realisht emrin. Kjo mund t\u00eb mendohet n\u00eb m\u00ebnyra t\u00eb ndryshme, por nuk \u00ebsht\u00eb e qart\u00eb se si \u00ebsht\u00eb m\u00eb mir\u00eb: si nj\u00eb problem teknik n\u00eb SQL, si nj\u00eb gabim n\u00eb l\u00ebshimin e Oracle, apo si nj\u00eb d\u00ebshtim i institucioneve burokratike? Sa e v\u00ebshtir\u00eb \u00ebsht\u00eb t\u00eb gjesh nj\u00eb gabim n\u00eb procesin q\u00eb \u00e7oi n\u00eb at\u00eb q\u00eb sistemi nuk ka nj\u00eb funksion si korrigjimi i gabimeve? A ishte e mundur t\u00eb kuptohej kjo para se njer\u00ebzit t\u00eb fillonin t\u00eb ankoeshin? <\/p>\n<p><\/p>\n<p>Nj\u00eb ilust\u00ebrim edhe m\u00eb t\u00eb qart\u00eb t\u00eb k\u00ebsaj \u00e7\u00ebshtjeje jan\u00eb historit\u00eb kur shofer\u00ebt p\u00ebrfundojn\u00eb n\u00eb lumenj p\u00ebr shkak t\u00eb t\u00eb dh\u00ebnave t\u00eb vjetra n\u00eb naviguar. Mir\u00eb, hartat duhet t\u00eb p\u00ebrdit\u00ebsohen vazhdimisht. Por sa faj ka TomTom p\u00ebr at\u00eb q\u00eb makina juaj p\u00ebrfundon n\u00eb det?<\/p>\n<p><\/p>\n<p>Po e them k\u00ebt\u00eb, q\u00eb po, \u2013 parap\u00eblqimi i m\u00ebsimit t\u00eb makinave do t\u00eb krijoj\u00eb probleme. Por k\u00ebto probleme do t\u00eb jen\u00eb t\u00eb ngjashme me ato me t\u00eb cilat jemi p\u00ebrballur n\u00eb t\u00eb kaluar\u00ebn, dhe ato do t\u00eb mund t\u00eb v\u00ebrehen dhe t\u00eb zgjidhen (ose jo) me forc\u00ebn q\u00eb e kemi arritur deri tani. Prandaj, skenari n\u00eb t\u00eb cilin parap\u00eblqimi i AI do t\u00eb d\u00ebmtoj\u00eb \u00ebsht\u00eb pak e mundur q\u00eb t\u00eb ndodh\u00eb me k\u00ebrkuesit kryesor\u00eb q\u00eb punojn\u00eb n\u00eb nj\u00eb organizat\u00eb t\u00eb madhe. \u00c7do gjas\u00eb \u00ebsht\u00eb q\u00eb ndonj\u00eb kontraktor teknologjik i pavler\u00eb ose shit\u00ebs programi do t\u00eb shkruaj\u00eb di\u00e7ka n\u00eb m\u00ebnyr\u00eb ad-hoc, duke p\u00ebrdorur komponente, biblioteka dhe mjete open-source q\u00eb ai nuk i kupton. Dhe klienti i pafat do t\u00eb mashtrohet nga fraza \u201cinteligjenca artificiale\u201d n\u00eb p\u00ebrshkrimin e produktit dhe, pa b\u00ebr\u00eb pyetje t\u00eb panevojshme, do t'i jap\u00eb at\u00eb punonj\u00ebsve t\u00eb tij me paga t\u00eb ulta, duke i nd\u00ebshkuar ata t\u00eb b\u00ebjn\u00eb at\u00eb q\u00eb thot\u00eb AI. Pik\u00ebrisht kjo ndodhi me bazat e t\u00eb dh\u00ebnave. Kjo nuk \u00ebsht\u00eb nj\u00eb problem i inteligjenc\u00ebs artificiale, dhe askush nuk e shpjegon probleme t\u00eb softuerit. Ky \u00ebsht\u00eb faktori njer\u00ebzor.<\/p>\n<p><\/p>\n<h2 id=\"zaklyuchenie\">P\u00ebrfundim<\/h2>\n<p><\/p>\n<blockquote><p><em>M\u00ebsimi i makinave mund t\u00eb b\u00ebj\u00eb gjith\u00e7ka q\u00eb mund ta m\u00ebsoni nj\u00eb qen, \u2013 por kurr\u00eb nuk mund t\u00eb jeni t\u00eb sigurt se \u00e7far\u00eb me t\u00eb v\u00ebrtet\u00eb e keni m\u00ebsuar at\u00eb qen.<\/em><\/p><\/blockquote>\n<p>M\u00eb duket shpesh se termi \u00abinteligjenc\u00eb artificiale\u00bb vet\u00ebm sa e v\u00ebshtir\u00ebson diskutimin p\u00ebr tema t\u00eb tilla. Ky term krijon nj\u00eb p\u00ebrshtypje t\u00eb gabuar se ne n\u00eb t\u00eb v\u00ebrtet\u00eb e kemi krijuar at\u00eb \u2014 k\u00ebt\u00eb inteligjenc\u00eb. Q\u00eb ne jemi n\u00eb rrug\u00eb drejt HAL9000 ose Skynet \u2014 drejt di\u00e7kaje q\u00eb n\u00eb t\u00eb v\u00ebrtet\u00eb <em>kupton<\/em>. Por jo. K\u00ebto jan\u00eb thjesht makina, dhe \u00ebsht\u00eb shum\u00eb m\u00eb e sakt\u00eb t'i krahasojm\u00eb, t\u00eb themi, me nj\u00eb lavatri\u00e7e. Ajo b\u00ebn shum\u00eb m\u00eb mir\u00eb pun\u00ebn e larjes se sa nj\u00eb njeri, por n\u00ebse ju hedhni en\u00ebt n\u00eb t\u00eb vend t\u00eb rrobave, ajo do t'i\u2026 laj\u00eb. En\u00ebt do t\u00eb b\u00ebhen madhe t\u00eb pastra. Por kjo nuk do t\u00eb jet\u00eb ajo q\u00eb prisnit, dhe kjo do t\u00eb ndodh\u00eb jo p\u00ebr shkak se sistemi ka disa paragjykime p\u00ebr en\u00ebt. Lavatri\u00e7a nuk e di as \u00e7far\u00eb jan\u00eb en\u00ebt, as \u00e7far\u00eb jan\u00eb rrobat \u2014 \u00ebsht\u00eb thjesht nj\u00eb shembull i automatizimit, konceptualisht i nj\u00ebjt\u00eb me at\u00eb si\u00e7 jan\u00eb automatizuar proceset m\u00eb par\u00eb.<\/p>\n<p><\/p>\n<p>Pavar\u00ebsisht se p\u00ebr \u00e7far\u00eb flitet, \u2014 p\u00ebr makina, aeroplan\u00eb ose databaza \u2014 k\u00ebto sisteme do t\u00eb jen\u00eb nj\u00ebkoh\u00ebsisht shum\u00eb t\u00eb fuqishme dhe shum\u00eb t\u00eb kufizuara. Ato do t\u00eb varen plot\u00ebsisht nga m\u00ebnyra se si njer\u00ebzit p\u00ebrdorin k\u00ebto sisteme, nga q\u00ebllimet e tyre t\u00eb mira ose t\u00eb k\u00ebqija dhe nga sa mir\u00eb e kuptojn\u00eb pun\u00ebn e tyre.<\/p>\n<p><\/p>\n<p>Prandaj, t\u00eb thuash se \u00abinteligjenca artificiale \u00ebsht\u00eb matematik\u00eb, prandaj nuk mund t\u00eb ket\u00eb paragjykime\u00bb \u00ebsht\u00eb plot\u00ebsisht e gabuara. Por po aq e gabuar \u00ebsht\u00eb t\u00eb thuash se m\u00ebsimi nga makinat \u00ebsht\u00eb \u00absubjektiv nga natyra\u00bb. M\u00ebsimi i makinave gjen modele n\u00eb t\u00eb dh\u00ebna, dhe cilat modele do t\u00eb gjej\u00eb tende p\u00ebr t\u00eb dh\u00ebnat, varet nga t\u00eb dh\u00ebnat, dhe t\u00eb dh\u00ebnat varen nga ne. Ashtu si\u00e7 dhe ajo q\u00eb ne b\u00ebjm\u00eb me to. M\u00ebsimi i makinave n\u00eb fakt b\u00ebn disa gj\u00ebra shum\u00eb m\u00eb mir\u00eb se ne \u2014 por qen\u00ebt, p\u00ebr shembull, jan\u00eb shum\u00eb m\u00eb efikas n\u00eb zbulimin e drog\u00ebs se sa njer\u00ebzit, q\u00eb nuk \u00ebsht\u00eb nj\u00eb arsye p\u00ebr t'i marr\u00eb ata si d\u00ebshmitar\u00eb dhe p\u00ebr t\u00eb marr\u00eb vendime t\u00eb bazuara tek deklaratat e tyre. Dhe qen\u00ebt, p\u00ebr t'u p\u00ebrmendur, jan\u00eb shum\u00eb m\u00eb t\u00eb men\u00e7ur se \u00e7do sistem m\u00ebsimi nga makinat.<\/p>\n<p><strong>P\u00ebrkthimi:<\/strong>\u00a0<noindex><a rel=\"nofollow\" href=\"https:\/\/habr.com\/en\/users\/757NF\/\">Diana Letcka<\/a><\/noindex>.<br \/>\n<strong>Redaktimi:<\/strong>\u00a0<noindex><a rel=\"nofollow\" href=\"http:\/\/teleg.one\/ponchiknews\">Aleksej Ivanov<\/a><\/noindex>.<br \/>\n<strong>Komuniteti:<\/strong> <noindex><a rel=\"nofollow\" href=\"http:\/\/teleg.one\/ponchiknews\">@PonchikNews<\/a><\/noindex>.<\/p>\n<p>Burimi: <a content=\"nofollow\" rel=\"nofollow\" href=\"https:\/\/habr.com\/ru\/post\/449224\/\">habr.com<\/a><\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"<p>tl;dr: \u041c\u0430\u0448\u0438\u043d\u043d\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u0438\u0449\u0435\u0442 \u0437\u0430\u043a\u043e\u043d\u043e\u043c\u0435\u0440\u043d\u043e\u0441\u0442\u0438 \u0432 \u0434\u0430\u043d\u043d\u044b\u0445. \u041d\u043e \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u044b\u0439 \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442 \u043c\u043e\u0436\u0435\u0442 \u0431\u044b\u0442\u044c \u00ab\u043f\u0440\u0435\u0434\u0432\u0437\u044f\u0442\u00bb \u2014\u00a0\u0442\u043e \u0435\u0441\u0442\u044c, \u043d\u0430\u0445\u043e\u0434\u0438\u0442\u044c \u043d\u0435\u0432\u0435\u0440\u043d\u044b\u0435 \u043f\u0430\u0442\u0442\u0435\u0440\u043d\u044b. \u041a \u043f\u0440\u0438\u043c\u0435\u0440\u0443, \u0441\u0438\u0441\u0442\u0435\u043c\u0430 \u043e\u0431\u043d\u0430\u0440\u0443\u0436\u0435\u043d\u0438\u044f \u0440\u0430\u043a\u0430 \u043a\u043e\u0436\u0438 \u043f\u043e \u0444\u043e\u0442\u043e\u0433\u0440\u0430\u0444\u0438\u0438 \u043c\u043e\u0436\u0435\u0442 \u043e\u0431\u0440\u0430\u0449\u0430\u0442\u044c \u043e\u0441\u043e\u0431\u043e\u0435 \u0432\u043d\u0438\u043c\u0430\u043d\u0438\u0435 \u043d\u0430 \u0441\u043d\u0438\u043c\u043a\u0438, \u0441\u0434\u0435\u043b\u0430\u043d\u043d\u044b\u0435 \u0432\u043e \u0432\u0440\u0430\u0447\u0435\u0431\u043d\u043e\u043c \u043a\u0430\u0431\u0438\u043d\u0435\u0442\u0435. \u041c\u0430\u0448\u0438\u043d\u043d\u043e\u0435 \u043e\u0431\u0443\u0447\u0435\u043d\u0438\u0435 \u043d\u0435 \u0443\u043c\u0435\u0435\u0442 \u043f\u043e\u043d\u0438\u043c\u0430\u0442\u044c: \u0435\u0433\u043e \u0430\u043b\u0433\u043e\u0440\u0438\u0442\u043c\u044b \u043b\u0438\u0448\u044c \u0432\u044b\u044f\u0432\u043b\u044f\u044e\u0442 \u0437\u0430\u043a\u043e\u043d\u043e\u043c\u0435\u0440\u043d\u043e\u0441\u0442\u0438 \u0432 \u0447\u0438\u0441\u043b\u0430\u0445, \u0438 \u0435\u0441\u043b\u0438 \u0434\u0430\u043d\u043d\u044b\u0435 \u043d\u0435 \u0440\u0435\u043f\u0440\u0435\u0437\u0435\u043d\u0442\u0430\u0442\u0438\u0432\u043d\u044b, \u0442\u0430\u043a\u0438\u043c \u0431\u0443\u0434\u0435\u0442 [&hellip;]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":1,"featured_media":24558,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-32781","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news"],"aioseo_notices":[],"aioseo_head":"\n\t\t<!-- All in One SEO 5.0.2 - aioseo.com -->\n\t<meta name=\"description\" content=\"tl;dr: .\" \/>\n\t<meta name=\"robots\" content=\"max-image-preview:large\" \/>\n\t<meta name=\"author\" content=\"Yuri Gagarin\"\/>\n\t<link rel=\"canonical\" href=\"https:\/\/prohoster.info\/sq\/blog\/news\/o-predvzyatosti-iskusstvennogo-intellekta\" \/>\n\t<meta name=\"generator\" content=\"All in One SEO (AIOSEO) 5.0.2\" \/>\n\t\t<meta property=\"og:locale\" content=\"sq_AL\" \/>\n\t\t<meta property=\"og:site_name\" content=\"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b\" \/>\n\t\t<meta property=\"og:type\" content=\"article\" \/>\n\t\t<meta property=\"og:title\" content=\"\ud83e\udd47\u041e \u043f\u0440\u0435\u0434\u0432\u0437\u044f\u0442\u043e\u0441\u0442\u0438 \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430 | ProHoster\" \/>\n\t\t<meta property=\"og:description\" content=\"tl;dr: .\" \/>\n\t\t<meta property=\"og:url\" content=\"https:\/\/prohoster.info\/sq\/blog\/news\/o-predvzyatosti-iskusstvennogo-intellekta\" \/>\n\t\t<meta property=\"og:image\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:secure_url\" content=\"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg\" \/>\n\t\t<meta property=\"og:image:width\" content=\"350\" \/>\n\t\t<meta property=\"og:image:height\" content=\"350\" \/>\n\t\t<meta property=\"article:published_time\" content=\"2019-10-31T18:48:54+00:00\" \/>\n\t\t<meta property=\"article:modified_time\" content=\"2019-10-31T18:48:54+00:00\" \/>\n\t\t<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<meta property=\"article:author\" content=\"https:\/\/www.facebook.com\/prohoster\" \/>\n\t\t<!-- All in One SEO -->\n\n","aioseo_head_json":{"title":"\ud83e\udd47P\u00ebr paragjykimin e inteligjenc\u00ebs artificiale | ProHoster","description":"tl;dr: .","canonical_url":"https:\/\/prohoster.info\/sq\/blog\/news\/o-predvzyatosti-iskusstvennogo-intellekta","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"sq_AL","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47\u041e \u043f\u0440\u0435\u0434\u0432\u0437\u044f\u0442\u043e\u0441\u0442\u0438 \u0438\u0441\u043a\u0443\u0441\u0441\u0442\u0432\u0435\u043d\u043d\u043e\u0433\u043e \u0438\u043d\u0442\u0435\u043b\u043b\u0435\u043a\u0442\u0430 | ProHoster","og:description":"tl;dr: .","og:url":"https:\/\/prohoster.info\/sq\/blog\/news\/o-predvzyatosti-iskusstvennogo-intellekta","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2019-10-31T18:48:54+00:00","article:modified_time":"2019-10-31T18:48:54+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":{"post_id":"32781","title":null,"description":null,"keywords":null,"keyphrases":null,"primary_term":null,"canonical_url":null,"og_title":null,"og_description":null,"og_object_type":"default","og_image_type":"default","og_image_url":null,"og_image_width":null,"og_image_height":null,"og_image_custom_url":null,"og_image_custom_fields":null,"og_video":null,"og_custom_url":null,"og_article_section":null,"og_article_tags":null,"twitter_use_og":false,"twitter_card":"default","twitter_image_type":"default","twitter_image_url":null,"twitter_image_custom_url":null,"twitter_image_custom_fields":null,"twitter_title":null,"twitter_description":null,"schema":{"blockGraphs":[],"customGraphs":[],"default":{"data":{"Article":[],"Course":[],"Dataset":[],"FAQPage":[],"Movie":[],"Person":[],"Product":[],"ProductReview":[],"Car":[],"Recipe":[],"Service":[],"SoftwareApplication":[],"WebPage":[]},"graphName":"","isEnabled":true},"graphs":[]},"schema_type":null,"schema_type_options":null,"pillar_content":false,"robots_default":true,"robots_noindex":false,"robots_noarchive":false,"robots_nosnippet":false,"robots_nofollow":false,"robots_noimageindex":false,"robots_noodp":false,"robots_notranslate":false,"robots_max_snippet":null,"robots_max_videopreview":null,"robots_max_imagepreview":"large","priority":null,"frequency":null,"local_seo":null,"seo_analyzer_scan_date":"2026-01-21 12:30:22","breadcrumb_settings":null,"limit_modified_date":false,"reviewed_by":null,"ai":null,"created":"2021-03-01 02:52:26","updated":"2026-01-21 12:30:22","focus_keyword":null,"additional_keywords":null,"truseo_locale":null},"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/32781","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/comments?post=32781"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/posts\/32781\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/media\/24558"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/media?parent=32781"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/categories?post=32781"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/sq\/wp-json\/wp\/v2\/tags?post=32781"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}