N\u00eb fund t\u00eb vitit t\u00eb kaluar, kur kompania e njohur kineze e Inteligjenc\u00ebs Artificiale, DeepSeek, hodhi n\u00eb treg modelin e madh gjuh\u00ebsor me nj\u00eb \u00e7mim t\u00eb lir\u00eb, ajo p\u00ebrmbysi parashikimet e kahershme mbi t\u00eb ardhmen e Inteligjenc\u00ebs Artificiale, shkruan The Economist.<\/p>\n
Kjo \u00ebsht\u00eb e r\u00ebnd\u00ebsishme p\u00ebr k\u00ebdo q\u00eb del n\u00eb krye t\u00eb betej\u00ebs s\u00eb madhe globale p\u00ebr ep\u00ebrsi n\u00eb Inteligjenc\u00ebn Artificiale. Zhvilluesit tani po rishikojn\u00eb se sa harduer, energji dhe t\u00eb dh\u00ebna nevojiten. Nj\u00eb tjet\u00ebr faktor kontribuues dhe m\u00eb pak i diskutuar n\u00eb Inteligjenc\u00ebn Artificiale, \u00ebsht\u00eb fuqia pun\u00ebtore.<\/p>\n
P\u00ebr njeriun e zakonsh\u00ebm, Inteligjenca Artificiale ka t\u00eb b\u00ebj\u00eb vet\u00ebm me robot\u00eb, makineri dhe modele. \u00cbsht\u00eb nj\u00eb teknologji q\u00eb vret vendet e pun\u00ebs. N\u00eb fakt, miliona pun\u00ebtor\u00eb jan\u00eb t\u00eb p\u00ebrfshir\u00eb n\u00eb prodhimin e modeleve t\u00eb Inteligjenc\u00ebs Artificiale.<\/p>\n
Pjesa m\u00eb e madhe e pun\u00ebs s\u00eb tyre p\u00ebrfshin detyra si etiketimi i objekteve n\u00eb imazhet e rrug\u00ebve n\u00eb m\u00ebnyr\u00eb q\u00eb t\u00eb trajnohen makinat vet\u00ebdrejtuese apo etiketimi i fjal\u00ebve n\u00eb regjistrimet audio t\u00eb p\u00ebrdorura p\u00ebr t\u00eb trajnuar sistemet e njohjes s\u00eb t\u00eb folurit.<\/p>\n
Teknikisht, k\u00ebta pun\u00ebtor\u00eb t\u00eb quajtur \u201csh\u00ebnues\u201d, i japin t\u00eb dh\u00ebnave kontekstin q\u00eb u nevojitet kompjuter\u00ebve, p\u00ebr t\u00eb p\u00ebrpunuar lidhjet statistikore midis komponent\u00ebve t\u00eb nj\u00eb grupi t\u00eb dh\u00ebnash dhe t\u00eb kuptuarit e tyre nga qeniet njer\u00ebzore.<\/p>\n
N\u00eb fakt, kushdo q\u00eb ka kryer me sukses nj\u00eb test CAPTCHA, mund t\u00eb ket\u00eb ndihmuar pa dashje n\u00eb trajnimin e Inteligjenc\u00ebs Artificiale.<\/p>\n
Kjo \u00ebsht\u00eb pjesa \u201cjot\u00ebrheq\u00ebse\u201d e industris\u00eb, si\u00e7 shprehet Alex Wang, drejtuesi i firm\u00ebs s\u00eb t\u00eb dh\u00ebnave Scale AI.<\/p>\n
Megjith\u00ebse Scale AI thot\u00eb se pjesa m\u00eb e madhe e pun\u00ebs s\u00eb kontribuuesve t\u00eb saj kryhet n\u00eb Amerik\u00eb dhe Europ\u00eb, n\u00eb industrin\u00eb n\u00eb t\u00ebr\u00ebsi, pjesa m\u00eb e madhe e pun\u00ebs u jepet njer\u00ebzve n\u00eb vendet e varfra t\u00eb bot\u00ebs, ku shum\u00eb persona t\u00eb arsimuar k\u00ebrkojn\u00eb pun\u00eb.<\/p>\n
Qeveria kineze ka bashkuar forcat me kompanit\u00eb e teknologjis\u00eb, si Alibaba dhe JD.com, p\u00ebr t\u00eb pun\u00ebsuar njer\u00ebz t\u00eb till\u00eb n\u00eb pjes\u00eb t\u00eb larg\u00ebta t\u00eb vendit.<\/p>\n
N\u00eb Indi, kompania Nasscom, llogarit se t\u00eb ardhurat nga \u201csh\u00ebnuesit\u201d mund t\u00eb arrijn\u00eb 7 miliard\u00eb dollar\u00eb n\u00eb vit dhe mund t\u00eb pun\u00ebsohen 1 milion njer\u00ebz n\u00eb vend deri n\u00eb vitin 2030.<\/p>\n
Kjo \u00ebsht\u00eb dometh\u00ebn\u00ebse, sepse e gjith\u00eb industria e teknologjis\u00eb s\u00eb informacionit n\u00eb Indi vlen 254 miliard\u00eb dollar\u00eb n\u00eb vit (p\u00ebrfshir\u00eb harduerin) dhe pun\u00ebson 5.5 milion\u00eb njer\u00ebz.<\/p>\n
\u201cSh\u00ebnuesit\u201d jan\u00eb krahasuar prej koh\u00ebsh me prind\u00ebrit, duke st\u00ebrvitur modelet e Inteligjenc\u00ebs Artificiale dhe duke i ndihmuar ata t\u00eb kuptojn\u00eb bot\u00ebn. Por modelet e fundit nuk kan\u00eb nevoj\u00eb p\u00ebr udh\u00ebzime n\u00eb t\u00eb nj\u00ebjt\u00ebn m\u00ebnyr\u00eb. Nd\u00ebrsa teknologjia po rritet, a do t\u00eb b\u00ebhen m\u00ebsuesit e saj t\u00eb tep\u00ebrt?<\/p>\n
Sh\u00ebnimi i t\u00eb dh\u00ebnave nuk \u00ebsht\u00eb asgj\u00eb e re. Fei Fei Li, nj\u00eb shkenc\u00ebtare kompjuterike amerikane e njohur si \u201ckumbara e Inteligjenc\u00ebs Artificiale\u201d, ka nxitur lindjen e k\u00ebsaj industrie n\u00eb mesin e viteve 2000, kur krijoi ImageNet, grupin m\u00eb t\u00eb madh t\u00eb t\u00eb dh\u00ebnave t\u00eb imazheve n\u00eb at\u00eb koh\u00eb.<\/p>\n
Zonja Li e kuptoi se n\u00ebse do t\u2019i paguante student\u00ebt e kolegjit p\u00ebr t\u00eb grupuar imazhet, kjo pun\u00eb do t\u00eb zgjaste 90 vjet. Prandaj, ajo pun\u00ebsoi njer\u00ebz anemban\u00eb bot\u00ebs, duke p\u00ebrdorur Mechanical Turk, nj\u00eb platform\u00eb n\u00eb internet e drejtuar nga Amazon.<\/p>\n
Ajo p\u00ebrfitoi rreth 3.2 milion\u00eb imazhe t\u00eb organizuara n\u00eb grupe t\u00eb dh\u00ebnash, n\u00eb dy vjet e gjysm\u00eb. S\u00eb shpejti, edhe laborator\u00eb t\u00eb tjer\u00eb t\u00eb Inteligjenc\u00ebs Artificiale filluan t\u00eb kontraktojn\u00eb pun\u00ebn n\u00eb k\u00ebt\u00eb m\u00ebnyr\u00eb.<\/p>\n
Me kalimin e koh\u00ebs, zhvilluesit u lodh\u00ebn me sh\u00ebnimet e cil\u00ebsis\u00eb s\u00eb ul\u00ebt t\u00eb b\u00ebra nga pun\u00ebtor\u00eb t\u00eb patrajnuar. U shfaq\u00ebn kompani t\u00eb t\u00eb dh\u00ebnave t\u00eb Inteligjenc\u00ebs Artificiale, si Sama dhe iMerit. Ato pun\u00ebsuan pun\u00ebtor\u00eb n\u00eb vendet e varfra.<\/p>\n
Puna joformale e \u201csh\u00ebnuesve\u201d vazhdoi, por u krijuan platforma t\u00eb posa\u00e7me p\u00ebr pun\u00ebn n\u00eb Inteligjenc\u00ebn Artificiale, si ato t\u00eb drejtuara nga firma Scale AI, e cila trajnon punonj\u00ebsit.<\/p>\n
Banka Bot\u00ebrore llogarit se midis 4.4% dhe 12.4% e fuqis\u00eb pun\u00ebtore globale, \u00ebsht\u00eb e p\u00ebrfshir\u00eb n\u00eb k\u00ebt\u00eb fush\u00eb. Krystal Kauffman, nj\u00eb banor n\u00eb Mi\u00e7igan, i cili ka nj\u00eb dekad\u00eb q\u00eb punon me t\u00eb dh\u00ebna t\u00eb tilla n\u00eb internet, mendon se kompanit\u00eb e teknologjis\u00eb kan\u00eb interes p\u00ebr ta mbajtur t\u00eb fshehur k\u00ebt\u00eb fuqi pun\u00ebtore.<\/p>\n
\u201cAto po shesin iden\u00eb magjike se t\u00eb gjitha k\u00ebto gj\u00ebra ndodhin vet\u00eb\u201d, thot\u00eb zonja Kauffman. \u201cPa pjes\u00ebn magjike, Inteligjenca Artificiale \u00ebsht\u00eb thjesht nj\u00eb tjet\u00ebr produkt\u201d.<\/p>\n
N\u00eb industri \u00ebsht\u00eb hapur nj\u00eb debat p\u00ebr trajtimin e pun\u00ebtor\u00ebve q\u00eb q\u00ebndrojn\u00eb pas Inteligjenc\u00ebs Artificiale. Kompanit\u00eb ngurrojn\u00eb t\u00eb b\u00ebjn\u00eb publike t\u00eb dh\u00ebnat mbi pagat. Por \u201csh\u00ebnuesit\u201d amerikan\u00eb n\u00eb p\u00ebrgjith\u00ebsi mendojn\u00eb se 10-20 dollar\u00eb ora \u00ebsht\u00eb nj\u00eb pages\u00eb e mir\u00eb n\u00eb platformat online.<\/p>\n
Ata q\u00eb punojn\u00eb n\u00eb vendet e varfra, shpesh marrin 4-8 dollar\u00eb ora. N\u00eb shum\u00eb raste, ka mjete v\u00ebzhgimi q\u00eb ndjekin veprimtarin\u00eb kompjuterike t\u00eb pun\u00ebtor\u00ebve dhe d\u00ebnohen n\u00ebse punojn\u00eb ngadal\u00eb.<\/p>\n
Firma Scale AI \u00ebsht\u00eb paditur disa her\u00eb p\u00ebr praktikat e pun\u00ebsimit. Firma i ka mohuar akuzat dhe thot\u00eb: \u201cNe planifikojm\u00eb t\u00eb mbrohemi fuqish\u00ebm\u201d.<\/p>\n
Por \u00e7\u00ebshtja m\u00eb e madhe, \u00ebsht\u00eb se puna e \u201csh\u00ebnuesve\u201d po thahet. Deri diku, kjo ishte e pashmangshme.<\/p>\n
N\u00ebse Inteligjenca Artificiale dikur ishte nj\u00eb f\u00ebmij\u00eb q\u00eb kishte nevoj\u00eb p\u00ebr nj\u00eb prind p\u00ebr t\u2019i v\u00ebn\u00eb n\u00eb dukje gj\u00ebrat dhe p\u00ebr ta ndihmuar t\u00eb kuptonte bot\u00ebn p\u00ebrreth tij, tani teknologjia \u00ebsht\u00eb rritur n\u00eb nj\u00eb adoleshent q\u00eb ka nevoj\u00eb p\u00ebr udh\u00ebzime dhe k\u00ebshilla t\u00eb rast\u00ebsishme.<\/p>\n
Laborator\u00ebt e Inteligjenc\u00ebs Artificiale po p\u00ebrdorin gjithnj\u00eb e m\u00eb shum\u00eb t\u00eb dh\u00ebna t\u00eb para-etiketuara nga laborator\u00eb t\u00eb tjer\u00eb, t\u00eb cil\u00ebt p\u00ebrdorin algoritme p\u00ebr t\u00eb vendosur etiketime n\u00eb grupet e t\u00eb dh\u00ebnave.<\/p>\n
Marrim shembullin e traktor\u00ebve me vet\u00ebdrejtim t\u00eb nd\u00ebrtuar nga firma Blue River Technology, nj\u00eb deg\u00eb e kompanis\u00eb John Deere, gjigant i pajisjeve bujq\u00ebsore. Tre vjet m\u00eb par\u00eb, inxhinier\u00ebt amerikan\u00eb ngarkonin foto n\u00eb cloud p\u00ebr t\u2019i kaluar stafit n\u00eb Hubli, Indi, udh\u00ebzime t\u00eb kujdesshme se \u00e7far\u00eb duhej etiketuar: traktor\u00eb, nd\u00ebrtesa, pajisje vadit\u00ebse, etj.<\/p>\n
Tani zhvilluesit p\u00ebrdorin t\u00eb dh\u00ebna t\u00eb para-etiketuara. Ata ende kan\u00eb nevoj\u00eb p\u00ebr nj\u00eb staf p\u00ebr t\u00eb kontrolluar etiketimin dhe p\u00ebr t\u2019u marr\u00eb me \u201crastet e skajshme\u201d, p\u00ebr shembull kur nj\u00eb re pluhuri err\u00ebson nj\u00eb pjes\u00eb t\u00eb peizazhit ose nj\u00eb hije peme err\u00ebson t\u00eb korrat, duke ngat\u00ebrruar modelin. Por nj\u00eb proces q\u00eb m\u00eb par\u00eb zgjaste disa muaj, tani k\u00ebrkon disa jav\u00eb.<\/p>\n
Vala m\u00eb e fundit e modeleve t\u00eb Inteligjenc\u00ebs Artificiale, ka ndryshuar shum\u00eb pun\u00ebn me t\u00eb dh\u00ebnat. Q\u00eb nga viti 2022, kur firma OpenAI p\u00ebr her\u00eb t\u00eb par\u00eb e lejoi publikun t\u00eb p\u00ebrdorte chatbot-in e saj ChatGPT, ka pasur interes gjithnj\u00eb e m\u00eb t\u00eb madh p\u00ebr modelet e m\u00ebdha gjuh\u00ebsore.<\/p>\n
T\u00eb dh\u00ebnat nga firma k\u00ebrkimore Pitchbook tregojn\u00eb se financimi i kapitalit sip\u00ebrmarr\u00ebs p\u00ebr firmat startup t\u00eb Inteligjenc\u00ebs Artificiale n\u00eb bot\u00eb u rrit me m\u00eb shum\u00eb se 50% n\u00eb vitin 2024, duke arritur n\u00eb 131.5 miliard\u00eb dollar\u00eb, edhe pse fondet p\u00ebr firmat e tjera startup, ran\u00eb.<\/p>\n
Pjesa m\u00eb e madhe e k\u00ebtyre parave po shkon n\u00eb teknika t\u00eb reja p\u00ebr zhvillimin e Inteligjenc\u00ebs Artificiale, t\u00eb cilat nuk kan\u00eb nevoj\u00eb p\u00ebr sh\u00ebnimin e t\u00eb dh\u00ebnave n\u00eb t\u00eb nj\u00ebjt\u00ebn m\u00ebnyr\u00eb. Iva Gumnishka n\u00eb firm\u00ebn Humans in the Loop, thot\u00eb se kompanit\u00eb q\u00eb pun\u00ebsojn\u00eb \u201csh\u00ebnues\u201d me aft\u00ebsi t\u00eb ul\u00ebta, po \u201cmbeten pas\u201d.<\/p>\n
Ka ende k\u00ebrkes\u00eb p\u00ebr \u201csh\u00ebnuesit\u201d, por puna e tyre ka ndryshuar. Nd\u00ebrsa bizneset po fillojn\u00eb t\u00eb p\u00ebrdorin Inteligjenc\u00ebn Artificiale, ata po nd\u00ebrtojn\u00eb modele m\u00eb t\u00eb vogla t\u00eb posa\u00e7me dhe po k\u00ebrkojn\u00eb \u201csh\u00ebnues\u201d me arsim t\u00eb lart\u00eb. \u00cbsht\u00eb b\u00ebr\u00eb e zakonshme q\u00eb reklamat p\u00ebr pun\u00ebsimin e \u201csh\u00ebnuesve\u201d t\u00eb k\u00ebrkojn\u00eb nj\u00eb doktoratur\u00eb ose aft\u00ebsi n\u00eb kodim dhe shkenc\u00eb.<\/p>\n
Tani q\u00eb studiuesit po p\u00ebrpiqen ta b\u00ebjn\u00eb Inteligjenc\u00ebn Artificiale m\u00eb shum\u00ebgjuh\u00ebshe, po rritet edhe k\u00ebrkesa p\u00ebr \u201csh\u00ebnues\u201d q\u00eb flasin gjuh\u00eb t\u00eb tjera p\u00ebrve\u00e7 anglishtes. Sushovan Das, dentist q\u00eb punon n\u00eb projekte mjek\u00ebsore q\u00eb lidhen me Inteligjenc\u00ebn Artificiale, mendon se puna e \u201csh\u00ebnuesve\u201d nuk do t\u00eb zhduket kurr\u00eb.<\/p>\n
\u201cKjo bot\u00eb po zhvillohet vazhdimisht, \u2013 thot\u00eb ai. \u2013 Pra, Inteligjenca Artificiale do t\u00eb p\u00ebrmir\u00ebsohet pa pushim\u201d.<\/p>\n
Nga ana tjet\u00ebr, po shfaqen role t\u00eb reja p\u00ebr njer\u00ebzit, n\u00eb trajnimin e Inteligjenc\u00ebs Artificiale. Firma k\u00ebrkimore Epoch AI, llogarit se sasia e tekstit me cil\u00ebsi t\u00eb lart\u00eb t\u00eb disponuesh\u00ebm p\u00ebr trajnimin e modeleve mund t\u00eb shteroj\u00eb deri n\u00eb vitin 2026.<\/p>\n
Disa laborator\u00eb t\u00eb Inteligjenc\u00ebs Artificiale po pun\u00ebsojn\u00eb njer\u00ebz p\u00ebr t\u00eb shkruar copa teksti dhe rreshta kodi mbi t\u00eb cilat mund t\u00eb trajnohen modelet e Inteligjenc\u00ebs Artificiale.<\/p>\n
T\u00eb tjer\u00eb po blejn\u00eb t\u00eb dh\u00ebna sintetike, t\u00eb krijuara duke p\u00ebrdorur algoritme kompjuterike dhe po pun\u00ebsojn\u00eb njer\u00ebz p\u00ebr t\u2019i verifikuar ato. \u201cT\u00eb dh\u00ebnat sintetike duhet t\u00eb jen\u00eb t\u00eb dh\u00ebna t\u00eb mira\u201d, thot\u00eb Wendy Gonzalez, shefi i firm\u00ebs Sama, q\u00eb ka operacione n\u00eb Afrik\u00ebn Lindore.<\/p>\n
Roli tjet\u00ebr i pun\u00ebtor\u00ebve \u00ebsht\u00eb t\u00eb vler\u00ebsojn\u00eb rezultatet nga modelet e Inteligjenc\u00ebs Artificiale dhe t\u00eb ndihmojn\u00eb p\u00ebr t\u2019u dh\u00ebn\u00eb atyre form\u00eb. Fal\u00eb k\u00ebsaj, ChatGPT ishte n\u00eb gjendje t\u00eb performonte m\u00eb mir\u00eb se chatbot-et e m\u00ebparshme.<\/p>\n
Xiaote Zhu n\u00eb firm\u00ebn Scale AI jep si shembull disa detyra t\u00eb hapura q\u00eb kryhen n\u00eb platform\u00ebn Outlier t\u00eb firm\u00ebs, e cila u prezantua n\u00eb vitin 2023 p\u00ebr t\u00eb ndihmuar ekspert\u00ebt me trajnimin e Inteligjenc\u00ebs Artificiale.<\/p>\n
Pun\u00ebtor\u00ebve u paraqiten dy p\u00ebrgjigje nga nj\u00eb chatbot q\u00eb k\u00ebshillon nj\u00eb itinerar p\u00ebr pushime n\u00eb Maldive. Ata duhet t\u00eb zgjedhin se cil\u00ebn p\u00ebrgjigje parap\u00eblqejn\u00eb, ta vler\u00ebsojn\u00eb at\u00eb, t\u00eb shpjegojn\u00eb pse p\u00ebrgjigjja \u00ebsht\u00eb e mir\u00eb apo e keqe dhe m\u00eb pas ta rishkruajn\u00eb p\u00ebrgjigjen p\u00ebr ta p\u00ebrmir\u00ebsuar.<\/p>\n
Shembulli i zonj\u00ebs Zhu \u00ebsht\u00eb nj\u00eb rast i ve\u00e7ant\u00eb. Megjithat\u00eb, \u00ebsht\u00eb e v\u00ebrtet\u00eb se nd\u00ebrhyrjet e njer\u00ebzve jan\u00eb thelb\u00ebsore p\u00ebr t\u2019u siguruar q\u00eb Inteligjenca Artificiale \u00ebsht\u00eb e sigurt dhe etike.<\/p>\n
N\u00eb nj\u00eb dokument q\u00eb u publikua pas hedhjes n\u00eb treg t\u00eb ChatGPT n\u00eb vitin 2022, kompania OpenAI tha se kishte pun\u00ebsuar ekspert\u00eb \u201cp\u00ebr t\u00eb hetuar dhe p\u00ebr t\u00eb dh\u00ebn\u00eb reagime\u201d n\u00eb lidhje me modelet e saj.<\/p>\n
N\u00eb fund t\u00eb k\u00ebtij procesi, modeli u trajnua q\u00eb t\u00eb mos u p\u00ebrgjigjej disa k\u00ebrkesave, si p\u00ebr shembull, k\u00ebrkes\u00ebs p\u00ebr t\u00eb shkruar tekste q\u00eb synojn\u00eb t\u2019i bindin njer\u00ebzit q\u00eb t\u2019i bashkohen grupit terrorist al-Kaeda.<\/p>\n
Fluturimi jasht\u00eb foles\u00eb<\/p>\n
Zhvilluesit e Inteligjenc\u00ebs Artificiale duan q\u00eb ta heqin nevoj\u00ebn p\u00ebr k\u00ebt\u00eb lloj kontributi njer\u00ebzor. Studimet tregojn\u00eb se deri n\u00eb 80% e koh\u00ebs q\u00eb shpenzohet p\u00ebr zhvillimin e Inteligjenc\u00ebs Artificiale, shkon p\u00ebr pun\u00ebn e t\u00eb dh\u00ebnave.<\/p>\n
Naveen Rao n\u00eb firm\u00ebn e Inteligjenc\u00ebs Artificiale me emrin Databricks, thot\u00eb se do t\u00eb donte q\u00eb modelet t\u00eb m\u00ebsonin vet\u00eb, ashtu si\u00e7 do t\u00eb donte q\u00eb t\u00eb b\u00ebnin f\u00ebmij\u00ebt e tij. \u201cUn\u00eb dua t\u00eb nd\u00ebrtoj njer\u00ebz vet\u00eb-efikas\u00eb\u201d, thot\u00eb ai. \u201cDua q\u00eb ata t\u00eb ken\u00eb kuriozitetin dhe t\u00eb kuptojn\u00eb si t\u2019i zgjidhin problemet. Nuk dua t\u2019i ushqej me lug\u00eb n\u00eb \u00e7do hap\u201d.<\/p>\n
Ka shum\u00eb entuziaz\u00ebm p\u00ebr m\u00ebsimin e pambik\u00ebqyrur. Ky p\u00ebrfshin ushqyerjen e modeleve t\u00eb Inteligjenc\u00ebs Artificiale me t\u00eb dh\u00ebna t\u00eb paetiketuara dhe t\u00eb m\u00ebsuarit sipas stilit p\u00ebrforcues, q\u00eb p\u00ebrdor prov\u00ebn dhe gabimin p\u00ebr t\u00eb p\u00ebrmir\u00ebsuar vendimmarrjen e makineris\u00eb s\u00eb Inteligjenc\u00ebs Artificiale.<\/p>\n
Kompanit\u00eb e Inteligjenc\u00ebs Artificiale, p\u00ebrfshir\u00eb DeepMind e kompanis\u00eb Google, kan\u00eb trajnuar makineri t\u00eb Inteligjenc\u00ebs Artificiale p\u00ebr t\u00eb fituar n\u00eb loj\u00ebra si Go dhe shah, duke luajtur me miliona gara kund\u00ebr vetes dhe duke gjurmuar se cilat strategji funksionojn\u00eb, pa dh\u00ebn\u00eb asnj\u00eb ndihm\u00eb njer\u00ebzore. Por kjo qasje autodidakte nuk funksionon jasht\u00eb sfer\u00ebs s\u00eb matematik\u00ebs dhe shkenc\u00ebs, t\u00eb pakt\u00ebn tani p\u00ebr tani.<\/p>\n
T\u00eb pasionuarit e teknologjis\u00eb kudo n\u00eb bot\u00eb jan\u00eb mahnitur nga fakti se sa i lir\u00eb dhe efikas \u00ebsht\u00eb modeli kinez DeepSeek. Por atyre nuk u p\u00eblqen fakti se DeepSeek po trajnohet duke p\u00ebrdorur reagime t\u00eb krijuara nga kompjuter\u00ebt dhe jo nga njer\u00ebzit.<\/p>\n
Modeli e kishte t\u00eb v\u00ebshtir\u00eb t\u2019u p\u00ebrgjigjej pyetjeve t\u00eb hapura. \u201cDallimi \u00ebsht\u00eb se me loj\u00ebrat si Go dhe shah, rezultati i d\u00ebshiruar \u00ebsht\u00eb krejt i qart\u00eb: fito loj\u00ebn\u201d, thot\u00eb Phelim Bradley, bashk\u00ebthemelues i firm\u00ebs Prolific, nj\u00eb tjet\u00ebr kompani e t\u00eb dh\u00ebnave t\u00eb Inteligjenc\u00ebs Artificiale. \u201cKurse modelet e m\u00ebdha gjuh\u00ebsore jan\u00eb m\u00eb t\u00eb nd\u00ebrlikuara dhe m\u00eb t\u00eb gjera\u201d.<\/p>\n
Bradley, si shum\u00eb teknik\u00eb, mendon se duhet t\u00eb p\u00ebrfshihen m\u00eb tep\u00ebr njer\u00ebz n\u00eb trajnimin e Inteligjenc\u00ebs Artificiale, jo m\u00eb pak. Larmia e fuqis\u00eb pun\u00ebtore ka r\u00ebnd\u00ebsi.<\/p>\n
Kur ChatGPT u publikua disa vite m\u00eb par\u00eb, njer\u00ebzit vun\u00eb re se ai e p\u00ebrdorte s\u00eb tep\u00ebrmi fjal\u00ebn \u201cg\u00ebrmoj\u201d. Kjo fjal\u00eb u kthye n\u00eb nj\u00eb shenj\u00eb dalluese e Inteligjenc\u00ebs Artificiale, q\u00eb tregonte se nj\u00eb tekst i caktuar ishte shkruar nga nj\u00eb robot.<\/p>\n
N\u00eb fakt, \u201csh\u00ebnuesit\u201d n\u00eb Afrik\u00eb ishin pun\u00ebsuar p\u00ebr t\u00eb trajnuar modelin dhe fjala \u201cg\u00ebrmoj\u201d p\u00ebrdoret m\u00eb shpesh n\u00eb anglishten afrikane, sesa n\u00eb anglishten amerikane ose britanike.<\/p>\n
N\u00eb t\u00eb nj\u00ebjt\u00ebn m\u00ebnyr\u00eb si\u00e7 p\u00ebrcillen aft\u00ebsit\u00eb dhe njohurit\u00eb e pun\u00ebtor\u00ebve te modelet e Inteligjenc\u00ebs Artificiale, ashtu p\u00ebrcillet edhe fjalori. Si\u00e7 thot\u00eb shprehja, n\u00eb rritjen e nj\u00eb f\u00ebmije merr pjes\u00eb gjith\u00eb fshati. \/ Monitor<\/p>\n
The post The Economist: A do t\u00eb b\u00ebhen pun\u00ebtor\u00ebt t\u00eb tep\u00ebrt nd\u00ebrsa teknologjia zhvillohet?<\/a> appeared first on Lapsi.al<\/a>.<\/p>","protected":false},"excerpt":{"rendered":" N\u00eb fund t\u00eb vitit t\u00eb kaluar, kur kompania e njohur kineze e Inteligjenc\u00ebs Artificiale, DeepSeek, hodhi n\u00eb treg modelin e madh gjuh\u00ebsor me nj\u00eb \u00e7mim t\u00eb lir\u00eb, ajo p\u00ebrmbysi parashikimet e kahershme mbi t\u00eb ardhmen e Inteligjenc\u00ebs Artificiale, shkruan The Economist. Kjo \u00ebsht\u00eb e r\u00ebnd\u00ebsishme p\u00ebr k\u00ebdo q\u00eb del n\u00eb krye t\u00eb betej\u00ebs s\u00eb madhe […]<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-20199","post","type-post","status-publish","format-standard","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/posts\/20199","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/comments?post=20199"}],"version-history":[{"count":0,"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/posts\/20199\/revisions"}],"wp:attachment":[{"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/media?parent=20199"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/categories?post=20199"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/zennihost.com\/test\/wp-json\/wp\/v2\/tags?post=20199"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}