Riaan 
“U-Bard ulapha ngempela ukuze asize abantu bathuthukise ukukhiqiza kwabo, basheshise imibono yabo, futhi bavele bakhulise ilukuluku labo,” kusho u-Sissie Hsiao, iphini likamongameli kwa-Google futhi ongomunye wabaphathi abasebenza ku-Bard.
Le nkampani isalelwe yizinyanga eziningi ekukhipheni inguqulo yokuqala ye-chatbot yayo emphakathini, isikhathi esibucayi izimbangi ezisisebenzisile ukulungisa imikhiqizo yazo. I-OpenAI, isiqalo esithuthukise i-ChatGPT, ivumele abasebenzisi ukuthi bahlole inguqulo yayo kusukela ngoNovemba futhi ngesonto eledlule ikhiphe ubuchwepheshe bayo obusezingeni eliphezulu, i-GPT-4. IMicrosoft yethule ithuluzi elifanayo enjinini yayo yokusesha ye-Bing ngoFebhuwari.
Kepha i-OpenAI kanye ne-Microsoft’s AI chatbots, esekelwe kubuchwepheshe obufanayo, nabo benze amaphutha futhi bakhombisa ukuziphatha okuxakile, njenge-Bing ye-Microsoft. ezichaza ngokuthi “Sydney” nokungena ezingxabanweni ezinobutha nabantu ababuza imibuzo. Lokho kuveza izinkampani ekugxekeni ukuthi ubuchwepheshe abukulungele ukusetshenziswa kabanzi. I-Google ihlole i-AI yayo kubasebenzi iminyaka – lowo omunye unjiniyela wangaleso sikhathi, uBlake Lemoine, athi wayesenozwelo – kodwa wenze iphutha ngasohlangothini lokuqapha nomphakathi.
Noma kunjalo, ukunensa kwe-Google kubangele ukukhungatheka kwabanye abasebenzi, abathi inkampani ilahle ibhola ngobuhlakani bokwenziwa obukhiqizayo, igama elibanzi lobuchwepheshe obusebenzisa ama-algorithms anamandla. baqeqeshwe ezingxenyeni ezinkulu ze-inthanethi ukukhiqiza okuqukethwe kwasekuqaleni, kusuka embhalweni othusayo ofana nomuntu kuye kumsebenzi wobuciko ocacile.
Abanye basola ukuqala kancane kweGoogle ngokukhathazeka kokuthi ubuchwepheshe bungase bulimaze isithunzi senkampani uma bukhishwa ingakalungeli ngokugcwele ukusetshenziswa ngumphakathi, ngokusho kwabantu abajwayelene nezingxoxo zangaphakathi abakhulume ngokungadalulwa ukuze bacobelelane ngolwazi olungakaphumela obala. Abanye abaholi be-AI, njengososayensi omkhulu wezobunhloli beMeta uYann LeCun, bathi amathuluzi akhishwe izinkampani ezinkulu zezobuchwepheshe kubonakale kungathandeki kangako futhi anamandla kunalawo akhishwe abaqalisi abancane ngoba anezinsimbi zokuvikela eziningi zokuvikela imiphumela ecasulayo noma eyingozi.
Yize i-Google nezinye izinkampani zisebenze kwezobuchwepheshe iminyaka, futhi enye isayensi yekhompiyutha ngemuva kwemikhiqizo yakamuva yenziwa umcabango emashumini eminyaka edlule, ama-bots ayaqhubeka nokumangaza abadali bawo kanye nabantu abajwayelekile ngalokho abangakwenza.
“Izinto zihamba ngokushesha,” kusho uLarry Birnbaum, uprofesa wesayensi yamakhompiyutha eNyuvesi yaseNorthwestern University ogxile ku-AI. “Ngisho nabantu abasebenza kule mishini, abaningi babo bamangale ngamakhono abo.”
Ngesonto eledlule, i-OpenAI ikhishwe inguqulo yakamuva ye-ChatGPT, engachaza izigcawu futhi ichaze okwenzeka kuzo, ibonise “amakhono okucabanga athuthukile,” ngokusho kwabadali bayo. Abanye abathuthukisi be-AI bathi ubuhlakani bezinga lomuntu buseduze, kuyilapho abagxeki bobuchwepheshe bethi i-hype ifihla ubungozi bangempela namathuba obuchwepheshe.
Kunoma ikuphi, izinga lomsebenzi kanye nentshisekelo ku-AI esezingeni eliphezulu liphakeme kunanini ngaphambili, futhi ochwepheshe bakholelwa ukuthi i-AI izosusa imisebenzi, idale emisha, futhi iguqule indlela abantu abasebenzisa ngayo ubuchwepheshe ne-inthanethi.
I-CEO yakwa-Google ibize i-AI ngokuthi “ijule kakhulu kunomlilo noma ugesi” futhi izinkampani ze-Big Tech ezifana ne-Google zilwela ukugibela igagasi futhi ziqinisekise ukubusa kwazo ku-inthanethi akusatshiswa ubuchwepheshe obuphazamisayo. Manje, zihamba ngokushesha kunalokho ebezingase zibe nakho.
“I-ChatGPT isheshise ukunyakaza okuhamba phambili kulo lonke uhlelo lwendalo, ishukumise abathengisi abaningi ukuthi babelane ngokukhululekile ngalokho abebekwenza isikhathi eside,” kusho uChirag Dekate, iphini likamongameli enkampanini yocwaningo lwezobuchwepheshe iGartner.
U-Eli Collins, iphini likamongameli we-Google kanye nesinye sezikhulu ezifaka isandla ku-Bard, uthe engxoxweni ukuthi inkampani isineminyaka isebenza nge-generative AI iminyaka nokuthi izinhlelo zayo ngeke zithonywe intshisekelo enkulu yakamuva esikhaleni.
“Sihamba kahle kakhulu ngale ndlela ebesikuyona, ethola ukuthi singakusebenzisa kanjani lokhu ngokuzibophezela ngezinga,” kusho yena.
I-Google sekuyisikhathi eside iyinkosi yokusesha kanye ne-AI, ichitha izigidigidi zamaRandi ekuthuthukiseni ubuchwepheshe obusezingeni eliphezulu. Enqubeni ngemuva kwempumelelo, isiqhwaga sezobuchwepheshe sihlanganise lezo ntuthuko emikhiqizweni yaso ekhona – kusukela enjinini yaso yokusesha kuya kumamephu amathuluzi okuhumushangenkathi futhi kushicilela amaphepha esayensi kwezinye zentuthuko.
Sekuyiminyaka, onjiniyela ngaphakathi enkampanini beveza ukuthi amamodeli ezilimi ezinkulu, ubuchwepheshe obusekela ama-chatbots afana ne-Bard ne-ChatGPT, angasetshenziswa ekusesheni ukuphendula imibuzo ngokuqondile, kunokuhlinzeka ngezixhumanisi noma amazwibela avela kwamanye amawebhusayithi, ngokusho kwabantu abajwayele. izingxoxo zangaphakathi.
Inkampani yenza ingxenye enkulu yemali engenayo ngezikhangiso zosesho, futhi abanye abasebenzi baphikise ngokuthi ama-chatbots ayinguquko elandelayo yomkhiqizo oyinhloko we-Google, kusho abantu. I-Bard isekelwe ku-LaMDA, imodeli yolimi enkulu inkampani eyaqala ukuyikhombisa ngoMeyi 2021.
Ekwindla ka-2021, i-Google ibingakaqiniseki ukuthi yimuphi umkhiqizo i-LaMDA ezofakwa kuwo kuqala, ngokusho kukaLemoine, owayengunjiniyela we-Google waxoshwa ngemuva kokutshela iWashington Post ukuthi ukholelwa ukuthi i-bot yayinomqondo. I-Search kanye nomsizi wezwi we-Google yibo ababezokhethwa kakhulu, kusho uLemoine.
I-Google isesimweni esihle sokusebenzisa ithuba lokuqhuma kwentshisekelo ku-AI, kusho uDekate, umhlaziyi wezobuchwepheshe. Le nkampani itshale imali iminyaka kumaseva amakhompyutha angakwazi ukwenza izibalo ezinkulu ezidingekayo ku-AI ethuthukisiwe, futhi ukuhlinzeka ngobuchwepheshe kwezinye izinkampani ezifuna ukubusebenzisa emikhiqizweni yabo kuzoba ibhizinisi elikhulu, esho.
“Empeleni imakethe ibheke lapho i-Google inethezeke ngendlela emangalisayo,” kusho uDekate. “Umthelela wangempela walokhu uzoba ku-ecosystem yozakwethu.”
Noma kunjalo, ukudideka kusasele phakathi kwabasebenzi mayelana nokuthi uBard kufanele abe yini. Ngesikhathi sokuboniswa kuka-Bard emcimbini wokusesha wango-Feb. 8, u-Prabhakar Raghavan, omunye wezikhulu eziphezulu ze-Google, wasebenzisa u-Bard ukuphendula imibuzo efana naleyo ebhalwa njalo ekusesheni. Ngaleso sikhathi, uthe ubuchwepheshe be-AI obukhiqizayo buzosetshenziselwa ukunikeza izimpendulo ezinde zemibuzo inkampani enquma ukuthi “ayikho impendulo efanele.” Inkampani ibonise imiphumela yosesho esikrinini.
Kepha ehholo ledolobha labasebenzi, isikhulu sithe uBard ubengahloselwe ukusetshenziswa njengethuluzi lokusesha – into abaphathi abayishilo kuleli sonto.
“U-Bard ungumhlanganyeli ohlakaniphile futhi kuyahambisana nokusesha,” kusho u-Hsiao.
Ngaphansi kwempendulo ka-Bard ngayinye, kuvela inkinobho evumela abantu ukuthi bashiye i-Bard ngokuchofoza futhi babuze umbuzo wabo esikhundleni kusesho lwe-Google.
Ama-Chatbots avela kubaqhudelani be-Google asesetshenziswa izigidi zabantu ukubhala izindatshana nemibiko, isikrini abangaba abalingani bothando bese ubhala ikhodi yekhompyutha. Izithombe ezenziwe ngamathuluzi e-AI ezifana ne-OpenAI’s DALL-E kanye ne-Stability AI’s Stable Diffusion ziye zagcwala izinkundla zokuxhumana izinyanga, ukudweba ukumangala nokumangala, ngenkathi iqala izinkulumo-mpikiswano ngokuthi kusho ukuthini ukuba yiciko nokuthi imithetho ye-copyright kufanele isetshenziswe kanjani kokuqukethwe okukhiqizwa yi-AI. Onxiwankulu bama-Venture bathela izigidigidi zamaRandi ekuqaleni kwe-AI, njengabo bonke abanye ukudilizwa kwe-tech funding ecosystem.
I-Google Bard ifana kakhulu ne-bots ekhishwe yi-Microsoft ne-OpenAI. Abasebenzisi bathayipha umbuzo ebhokisini lombhalo, futhi i-bot iphendula ngezimpendulo. Inani lemibuzo nezimpendulo zivaliwe ukuze kuvinjelwe i-bot ukuthi ithuthukiswe ekuthuthukiseni ubuntu bokulwa, njengoba kwenzeka ngo I-Microsoft’s chatbot ye-Bing ngemuva kokuthi abasebenzisi babe nezingxoxo zamahora amaningi nayo. Inkampani iphinde yavala ikhono lika-Bard lokukhiqiza ikhodi yekhompyutha, umkhawulo oyinhloko uma uqhathaniswa ne-ChatGPT.
Umdondoshiya wezobuchwepheshe uzoqala ukukhipha i-Bard ngoLwesibili kubantu ababhalise e-United States nase-United Kingdom. Ekugcineni, izotholakala emazweni amaningi nangezinye izilimi ngaphandle kwesiNgisi, kusho inkampani eposini le-blog.
UCollins noHsiao bathi inkampani izibophezele ekukhipheni imikhiqizo yayo ngokuzibophezela, futhi baphinde basho kaningi ngesikhathi bexoxisana ukuthi u-Bard usengumzamo. Idemo enikezwe izintatheli ngenhloso ifake isibonelo sokuthi u-Bard wenza iphutha ngokusungula igama lesayensi elisha lesitshalo lapho ephendula imibuzo emayelana nezitshalo zasendlini okulula ukuzinakekela.
