Riaan Amathuluzi e-Artificial intelligence (AI) abesematheni kusukela kwafika i-ChatGPT November ngonyaka odlule. I-ChatGPT yengqondo ye-OpenAI isivuse umjaho wezikhali we-AI obangela ukukhula okungakaze kubonwe embonini, njengoba izinkampani eziningi zigijima ukuze zikhiphe amathuluzi azo e-AI njenge-Google Bard kanye ne-Ernie Bot ye-Baidu.
Kodwa-ke, ukuthuthuka kwe-AI kubangele amagagasi ovalo nokungaqiniseki, ikakhulukazi mayelana namandla obuchwepheshe okushintsha imisebenzi yabantu.
Njengoba izindatshana ezingenakubalwa zisixwayisa kusengaphambili mayelana nokushabalala kwemisebenzi eminingi ezandleni ze-AI – 300 million imisebenzi yesikhathi esigcweleempeleni – okunganakwa phakathi kwalokhu kulandisa okuthusayo iqiniso lokuthi i-AI empeleni ingumkhiqizi okhiqiza imisebenzi emisha.
Omunye umsebenzi onjalo ovelile ngemuva kokuthuthuka kwe-AI “unjiniyela osheshayo” we-AI. Isibopho sabo esiyinhloko ukudala kanye nokwengamela imiyalo. Ngamanye amazwi, okokufaka, noma lokho oyala i-AI bot ukuthi ikwenze, okuqondisa izimpendulo nokuziphatha kwe-AI.
Njengokuvamile ubuchwepheshe obukhokha kakhulu imisebenzi, onjiniyela abasheshayo be-AI bangathola i- ngaphezulu kwama-US $335,000 (S$445,000) ngonyaka e-United States – nokho, ngokumangalisayo, ubunjiniyela bekhompyutha noma i-STEM degree ayisona isidingo sangaphambili sale ndima.
Kunalokho, indima idinga amakhono olimi aqinile angayala amamodeli e-AI – kungaba i-ChatGPT, BARD, noma i-Bing – ekukhiqizeni izimpendulo ezinembe kakhudlwana, ezihambisanayo, nezifaneleke ngokwemongo.
Imisebenzi yonjiniyela esheshayo ye-AI eSingapore
Uma kucatshangelwa ukuthi iSingapore ibeka a ukugcizelela okuphezulu ekuthuthukisweni kwe-AI, akumangazi ukuthi indima kanjiniyela esheshayo nayo iyadingeka kusifunda-dolobha. Ngemuva kwakho konke, imakethe ye-AI eSingapore kulindeleke ukuthi ifinyelele ku-US$16 billion (S$21.49 billion) ngo-2030.
Izinkampani eziningi ezivelele, okuhlanganisa i-Hepmil (inkampani engumzali ye-SGAG) kanye ne-ERA Real Estatesebeyamukele kakade i-AI yengxoxo njengendlela yokwenza izinqubo eziningi ngokuzenzakalelayo, kusukela esizukulwaneni se-imeyili kuya ekudaleni izithombe.
Ngisho nohulumeni waseSingapore unjalo njengamanje emisebenzini yokuhlanganisa I-ChatGPT ibe yi-Microsoft Word, ihlose “ukukhulula [civil] abaphathi bemisebenzi esezingeni eliphezulu”.
Ngokuzenzakalela imisebenzi efinyelelwa kalula, izinsiza zabantu ezibalulekile manje zingaqondiswa kabusha kweminye imisebenzi, okwenza ubunjiniyela obusheshayo bube yikhono elifunwa kakhulu.
Esikhathini esingangenyanga edlule, iNational University of Singapore (NUS) yathumela a ukuqasha ucingo ukufuna onjiniyela abasheshayo. Incazelo yomsebenzi ehlinzekwe yi-NUS igqamisa isibopho sokwenza imiyalo emisha noma amaketanga okwazisa ukuze kusetshenziswe amandla amamodeli ezilimi ezinkulu njenge-ChatGPT kanye ne-BARD ukuze enze inqwaba yemisebenzi ngempumelelo, okuhlanganisa ukukhiqiza izindaba namagama ezingoma.
Nakuba amaholo e-United States engafinyelela izibalo eziyisithupha, amaholo anikezwa izinkampani zaseSingapore anesizotha kakhulu. Isibonelo, i-NUS inikeza inkokhelo ye kuze kufike ku-S$96,000 ngonyaka ngendima.
Esikhundleni sokuqasha onjiniyela abasheshayo abazinikele, ezinye izinkampani zihlola ezinye izindlela ngokwandisa izibopho zabasebenzi nonjiniyela abakhona. Ngokwenza kanjalo, bathola ukonga ezindlekweni ezengeziwe zokuqasha.
Isibonelo, inkampani ehlala ndawonye i-HomeyDays ithumele a ukuqasha ucingo koMphathi Wezokuthengisa/Wokusebenza ongasebenzisa ngempumelelo ubuchwepheshe ukuze wenze lula futhi enze imisebenzi ehlukahlukene ibe lula.
Ukuvala igebe phakathi kokulindelwe ngabantu kanye nezimpendulo ze-AI
Kuleli qophelo, ungase uzibuze ukuthi kungani izinhlangano zifuna ukuqasha onjiniyela ngokushesha, naphezu kobuningi nokufinyeleleka kwama-chatbots e-AI atholakala mahhala. Noma ubani, nawo wonke umuntu, angaba umbhali ohloniphekile, umdwebi wezithombe kanye nomdwebi wopopayi ngobusuku obubodwa.
Kodwa-ke, iqiniso liwukuthi ukuzuza izimpendulo ezinembile nezifiselekayo kulawa mathuluzi e-AI kungaba inselele kakhulu.
Naphezu kwamakhono ayo, i-AI ayikwazi ukufunda ingqondo yakho. Kudinga ubunjiniyela bokwaziswa okucophelelayo, ukucwengwa, nokulungisa kahle ukuze kuthuthukiswe okuphumayo. Kungakho ukwenza ukwaziswa okusebenzayo nokunikeza imiyalelo ecacile kubalulekile ukuze uthole izimpendulo ezifiselekayo kumamodeli we-AI, okudinga ubuchwepheshe nomzamo omkhulu.
Eqinisweni, ngenxa yezinselelo ezihambisana nokuqhamuka nokwaziswa okuhle, izimakethe eziku-inthanethi ezifana I-Promptbase sekuvele, lapho onjiniyela abasheshayo bengathengisa imiyalelo yabo eklanywe ngokucophelela eklanyelwe amathuluzi e-AI ahlukene njenge-DALL-E ne-Midjourney.
Onjiniyela be-AI empeleni basebenza njengabalamuli, abanamakhono okuvala igebe phakathi kokulindelwe abantu kanye nezimpendulo ze-AI. Inani olithola kulawa mathuluzi e-AI lincike kuphela ekusebenzeni kokwaziswa kwakho.
Ngaphezu kwalokho, imiyalo idinga ukucutshungulwa ukuze ivumelane namamodeli e-AI ahlukene ukuze kuzuzwe imiphumela emihle. Amamodeli e-AI aklanywe ngezakhiwo ezithile futhi aqeqeshwa ezinhlotsheni ezithile zedatha, okusho ukuthi angase abe namandla ahlukahlukene, imikhawulo, namafomethi okokufaka athandwayo.
Ngenxa yalokho, ukwaziswa kobuciko okuhambisana namandla kanye nezidingo zemodeli ethile ye-AI kubalulekile ukuze kutholwe izimpendulo ezifiswayo, okuphinde kugqamise isidingo sonjiniyela be-AI.
Ngabe isidingo sonjiniyela be-AI sizohlala?
Yize abanye bengase baphikise ngokuthi isidingo sonjiniyela abasheshayo singase sinciphe kancane kancane njengoba i-AI ekhiqizayo iqhubeka nokuthuthuka, labo njiniyela basazodlala indima ebalulekile ekuqinisekiseni ukuthi lawa mamodeli ahlinzeka ngemiphumela engcono kakhulu.
Nakuba umgomo ekugcineni wezinhlelo ze-AI kuwukuba zibe lula futhi zikwazi ukukhiqiza okufiswayo ngokufaka okuncane, zisazodinga isiqondiso nesiqondiso ukuze kuqinisekiswe ukuthi imiphumela yazo ihambisana nokulindelwe abantu.
Ngaphezu kwalokho, amamodeli e-AI ahlukene anezakhiwo zawo, idatha yokuqeqesha, namafomethi okokufaka athandwayo. Onjiniyela abasheshayo bayawaqonda lawa ma-nuances futhi bangakwazi ukukhetha imiyalo eklanyelwe imodeli ngayinye ukuze bathole imiphumela emihle.
Bangakwazi ukuthuthukisa ulwazi lwabo namakhono ukuze bahlole amandla aphelele e-AI ekhiqizayo, bembule izinhlelo zokusebenza ezintsha futhi basebenzise amacala angaphezu kwalokho okungafinyelelwa amamodeli ngokuzimela.
Ngakho-ke, njengoba isidingo sonjiniyela abasheshayo singase sithuthuke njengoba ubuchwepheshe buqhubeka, indima yabo ekubumbeni nasekwenzeni kahle imiphumela yamamodeli akhiqizayo e-AI kungenzeka ihlale ibalulekile futhi ibalulekile.
Ikhredithi Yesithombe Esifakiwe: Izithombe ze-Getty
