Ngokombiko we-Futurism wakwamanye amazwe namuhla (mhla ziyi-12), uSimar Bajaj, ongumfundi wezokwelapha eStanford University School of Medicine futhi eyintatheli, kanye noJoseph Sakran, udokotela ohlinzayo wezingozi eJohns Hopkins Medical Center futhi oyisekela lomqondisi omkhulu wezokuhlinza, bakholelwa ukuthi izinkinga ezidalwa ukusetshenziswa kwe-AI ngabafundi bezokwelapha sezikhathaza kakhulu.
UBajaj noSakran babhala ku-The Guardian ukuthi ingozi yokusebenzisa i-AI ngabafundi bezokwelapha ayigcini nje ngokuncipha kwamakhono, kodwa ingase isho nokuthi lawo makhono awazange akheke ngempela kwasekuqaleni. “Udokotela okhohliwe ukuthi acabange ngendlela enengqondo usengakwazi ukuphinde alithole lelo khono, kodwa udokotela ongakaze afunde ukucabanga ngendlela enengqondo angase angakwazi.”
UBajaj noSakran bathi kunezinto ezimbili ezitholakele muva nje ezikhathaza kakhulu.
Okokuqala kuvela kudatha yangaphakathi. Cishe izingxenye ezimbili kwezintathu zodokotela eMelika seziyasebenzisa i-OpenEvidence, irobhothi lengxoxo le-AI eliklanyelwe odokotela abasebenza emtholampilo, okukhombisa ukuthi lobu buchwepheshe sebusungene kakhulu ezinqubweni zomsebenzi wezokwelapha futhi sebuye bahlanganiswa ngokujulile nomsebenzi wansuku zonke wabasebenzi bezempilo.
Okwesibili ucwaningo olusha olushicilelwe ku-Nature Medicine, lapho abacwaningi behlola khona ukwethembeka kwamamodeli amakhulu olimi akhethelwe ezokwelapha. Imiphumela yamangaza: ekuphenduleni imibuzo yezokwelapha, la mathuluzi e-AI ezokwelapha asebenza kabi kakhulu kunamarobhothi ezingxoxo asetshenziswa emikhakheni ehlukahlukene, afana ne-ChatGPT ne-Claude.
Ababhali bocwaningo bathi ukusebenza kwawo sekukonke kucishe kufane nokwe-“AI Overviews” yakwaGoogle. Phambilini, i-AI Overviews yayisidume kabi ngenxa yokungazinzi kokusebenza kwayo nezimpendulo ezinganembile.
Ngisho noma amathuluzi e-AI engasiza ekutholeni ulwazi, ingasaphathwa eyokuthi ukwethembeka kwawo kusenezinkinga ezinkulu, awakwazi ukuthatha indawo yokuthi umfundi acabange ngokwakhe. Lokhu kungenxa yokuthi ngesikhathi sokufunda, ukucabanga kanzima ngokwakho kuyiyona nto ebalulekile.
UBajaj noSakran babhala bathi: “Ngokwesibonelo, esikhathini esedlule lapho abaqeqeshwayo becelwa ukuba babhale uhlu lwezifo abangase bazixilonge, kwakudingeka bacabange kanzima, bagcine sebethule uhlu olungaphelele, bese befunda ezintweni abazishiyile, futhi ngezinye izikhathi bezizwe benamahloni ngenxa yalokho. Manje abaqeqeshwayo badinga nje ukubuza i-OpenEvidence ukuze bathole impendulo ecishe iphelele, ehlanganisa namathuba abangakaze bawacabange, ngaphandle kokudlula ehlazweni lokushiya ngaphandle isifo esingase sibe khona.”
UBajaj noSakran babhala bathi umehluko phakathi kwe-AI nentuthuko yangaphambilini yobuchwepheshe bezokwelapha ukuthi i-AI ayigcini nje ngokwandisa lokho odokotela abangakubona, kodwa iyangenelela ezindleleni zokucabanga okufanele zakheke ngesikhathi sokuqeqeshwa kwezokwelapha.
UBajaj noSakran baphakamise izindlela eziningana zokuxazulula le nkinga. Enye yazo ukuthi abaqeqeshwayo bacelwe ukuba baphathe amacala ngokuzimela ngezikhathi ezithile bengasizwa nhlobo yi-AI, okufana neseluleko se-Federal Aviation Administration yaseMelika sokuthi abashayeli bezindiza bavale uhlelo lokushayela ngokuzenzakalelayo ngezikhathi ezithile bese beshayela indiza ngezandla. Omunye umkhawulo onamandla ungase uqhamuke “emzweni wehlazo”.