I-Anthropic isanda kuthatha isinqumo sokufaka uphawu lwamanzi emiphumeleni ka-Claude. Lokhu kusho ukufaka ikhodi engabonakali emibhalweni ehleliwe ekhiqizwa yi-chatbot ukuze kubonakale ukuthi okuqukethwe kukhiqizwe ubuhlakani bokwenziwa.
Ngokokuqonda kwe-IT之家, i-Anthropic yethule le nqubomgomo entsha ukuze ihlangabezane nezimfuneko ze-Transparency Code ngaphansi kwe-European Union Artificial Intelligence Act (EU AI Act). Le mithetho manje idinga ukuthi izinkampani zobuchwepheshe zimake okuqukethwe okukhiqizwe noma okuhlelwe ubuhlakani bokwenziwa, futhi lolo phawu kumele lukwazi ukubonwa yizinhlelo zamakhompyutha. Nokho, nakuba abalawuli baseYurophu bengase baneliseke ngalokhu, abanye abasebenzisi be-AI babonakala bengajabule neze.
Ku-Reddit kulula ukubona ukwanda kokungeneliseki kwabasebenzisi, nakuba abanye abasebenzisi besayithi bengavumelani ngokuphelele nale mibono. Okunye kokuthunyelwe okunolaka kakhulu kuvela kumsebenzisi ogama lakhe lingu-visionode, i-akhawunti yakhe esineminyaka engaphansi kwamasonto amathathu ikhona. U-visionode ukholelwa ukuthi lolu hlelo olusha lokufaka uphawu lwamanzi luyitulo elinonya, elihlose ukwenza abasebenzisi abangenacala bama-chatbot emhlabeni wonke babe yizisulu.
Umbono oyinhloko ka-visionode ubonakala uwukuthi abasebenzisi abajwayelene no-Claude bangabhala kabusha okuqukethwe noma basebenzise ezinye izinsizakalo ze-AI ukucubungula imiphumela ka-Claude, ngaleyo ndlela bafihle izimpawu zokusebenzisa i-AI; kodwa abasebenzisi abavamile baka-Claude bona abanayo leyo nhlanhla.
“Ubani ozobanjwa? Nguwe. Umfundi osebenzisa u-Claude ukuze ahlele kabusha umbhalo; intatheli ecela i-AI ukuba ifingqe amarekhodi ezingxoxo angu-200 wamakhasi; umbhali obhekene nenkinga yokudala, ocela i-AI ukuba imnike omqondofana. Ngemva kokuqeda, bonke laba bantu bazosala benomdwebo wedijithali ebunzini.”
Lezi zibonelo empeleni azifaneleki ngokuphelele. Intatheli ecela i-AI ukuba ifingqe amarekhodi ezingxoxo angu-200 wamakhasi ngeke ithinteke ngoba nje isifinyezo sinophawu lwamanzi, ngaphandle uma ikopisha futhi inamathisele isifinyezo se-AI njengoba sinjalo esihlokweni sayo.
Ngokufanayo, uma umfundi ecela u-Claude ukuba “ahlehlele kabusha umbhalo” bese ekopisha futhi anamathisele ngqo okuqukethwe okukhiqizwe u-Claude ephepheni lakhe, nalokho akuhambisani nezimiso zokuziphatha kwezemfundo.
Abanye abasebenzisi be-Reddit nabo abazange bawusekele kakhulu umsindo walo mbhali. Omunye umsebenzisi wamane waphawula wathi: “Bhekani lo muntu.” Omunye wacela umbhali wokuthunyelwe kwasekuqaleni ukuba “aphefumule kakhulu.”
U-visionode akuyena yedwa okhalazayo. Omunye umsebenzisi owayenganelisekile wabiza la maphawu amanzi ngokuthi “awanasimilo” futhi “ayanyanyisa,” wathi yena wenza ingxenye enkulu yomsebenzi ngesikhathi esebenzisa u-Claude. Ngokombono wakhe, i-chatbot imane iy “ithuluzi” elimsiza ukwenza umsebenzi onzima.
“Nginikeze imiyalelo, ulwazi lwangemuva, izinqumo engazithatha kanye nokulungiswa okungenakubalwa; u-Claude uyithuluzi nje. Uma u-Claude eqala ukufaka uphawu lwamanzi kukhodi ayikhiqizayo noma kunoma yini enye, empeleni uzishayelani ihlombe?” kubuza lo msebenzisi.
Nokho, abanye abasebenzisi baphinde baphikisa lo mgxeki. “Akazishayeli ihlombe,” kuphendula omunye umsebenzisi, “lokhu kwenzelwa ukuthi okuqukethwe okukhiqizwe yi-AI kubonakale, ngoba okuqukethwe okukhiqizwe yi-AI kungadala ubungozi ezimweni ezihlukahlukene.”
Omunye umsebenzisi wamhleka usulu wathi: “Mfowethu, awukwazi ngisho nokugxeka u-Claude, kufanele ucele u-Claude akubhalele.”
Abanye abagxeki abazange basebenzise indaba “yezisulu,” kodwa babuza inqubomgomo entsha ye-Anthropic ngombono onemininingwane eminingi.
Ngokwesibonelo, omunye umsebenzisi ukholelwa ukuthi ukufaka uphawu lwamanzi kokuqukethwe okukhiqizwe yi-AI yona uqobo eqeqeshwe ngokumunca imisebenzi yabanye abantu ngobuningi kubonisa izinga elithile lokusebenzisa izindinganiso ezimbili.
“Ngicabanga ukuthi lena yindlela yokuthuthuka enobuqili obukhulu,” kusho lo msebenzisi. “Angisebenzisi u-Claude ukubhala noma yini, kodwa uma kubhekwa ukuthi mangaki amamodeli e-AI athuthukile aqeqeshwa ngokuthola imisebenzi yabanye abantu njengedatha yokuqeqesha, ukufaka i-AI uphawu lwamanzi emisebenzini yakho kunokuhlekisa okwesabekayo.”
Nokho, ngokujwayelekile, abasebenzisi babonakala besawusekela lo mshini wokufaka uphawu lwamanzi, bewubheka njengendlela enengqondo yokulandelela okuqukethwe okukhiqizwe ama-algorithm.
“Asikho isizathu esanele sokuchaza ukuthi kungani lokhu kungeyona into enhle,” kusho umsebenzisi kwenye ingxoxo. “Isizathu kuphela sokungafuni lo mshini ukuthi ufuna ukuqamba amanga kwabanye.”