Ochwepheshe be-AI baxoxa ngokuthi bahlanganisa kanjani i-ROBUST AI ekunakekelweni kwezempilo, kungani ukusebenzisana okuphathelene nokuphazamiseka kubalulekile, kanye namandla okukhiqiza ai ocwaningweni.
UFeifei Li noLloyd ONLINE banikeze amazwi okuvula eMbusweni wezeMpilo wezeMpilo eStanford University School of Medicine ngoMeyi 14. Steve Fish
Iningi labantu elibanjwe yi-Artificial Intelligence selibe nohlobo oluthile lwe- "Aha" umzuzu, ukuvula izingqondo zazo ezweni lamathuba. Ekuphakamiseni kwezeMpilo kwezempilo ku-Meyi 14, Lloyd Minor, MD, Dean of Stanford University School of Medicine nePhini Mongameli yezindaba zezokwelapha eStanford University, wabelana ngombono wakhe.
Lapho umuntu osemusha onelukuluku acelwa ukuba afingqese okutholakele kwakhe maqondana nendlebe yangaphakathi, waphendukela ebuhlakani bokufakelwa. "Ngibuze, 'Uyini Isifo Sokuphakama Kwe-Dehiskence Syndrome?' Abancane batshele cishe ababambiqhaza abangama-4,000. Endabeni yemizuzwana, kwavela izigaba eziningana.
"Bahle, bahle impela," kusho yena. "Ukuthi lolu lwazi lwahlanganiswa kube yincazelo emfushane, evame ukwenziwa kabusha kwalesi sifo. Lokhu kuyamangalisa impela. "
Iningi labelana ngenkululelo encane yomcimbi we-Half-Day, okwakuyinto ephumelelayo yohlelo lwezempilo, iphrojekthi eyethulwe yiStanford University School of Medicility kanye neStanford Institute for Human-Institute for Human-Institute for Human-Institute for the Orvififial Intelligence Human Intelligence ubuhlakani. ubuhlakani ocwaningweni lwe-biomedical, ezemfundo kanye nokunakekelwa kweziguli. Izikhulumi zihlolisisile ukuthi zisho ukuthini ukusebenzisa ubuhlakani bokufakelwa emithini ngendlela engalusizo kuphela kodokotela nososayensi, kodwa futhi esobala, inobulungiswa futhi ilingana neziguli.
"Sikholwa ukuthi lobu buchwepheshe obuthuthukisa amandla abantu," kusho uFei-Fei Li, uprofesa wesayensi yekhompyutha eStanford School of Engineering kanye nomqondisi we-Hai. Isizukulwane ngemuva kwesizukulwane, ubuchwepheshe obusha bungaqhamuka ekulandeleni okusha kwama-antibiotics kuya ekunciphiseni ukuhlukahluka kwezinto eziphilayo kanye nokuveza izingxenye ezifihliwe ze-biology eziyisisekelo, i-AI isheshisa ukutholakala kwesayensi. Kepha akukhona konke lokhu okuzuzisayo. "Zonke lezi zinhlelo zokusebenza zingaba nemiphumela engahlosiwe, futhi sidinga ososayensi bekhompyutha abakhulisa futhi basebenzise [ubuhlakani bokufakelwa] ababambe iqhaza, abasebenza nabo odokotela nasezithembisweni nasezithembisweni nasekupheleni," kusho yena. "Imizamo efana ne-Resourle khombisa ukuzibophezela kwethu kulokhu."
Ukuhlanganiswa kwezigaba ezintathu zemithi yaseStanford - Isikole sezokwelapha, ukunakekelwa kwempilo yaseStanford kanye neStanford University School of Stanford University - futhi ukuxhumana kwawo kwezinye izingxenye zeStanford University kubeke endaweni lapho kufakwa khona lapho kuthuthukiswa khona ukuhlakanipha okungekhona okwangempela. Ukuphathwa kanye nezinkinga zokuhlanganisa emkhakheni wezempilo kanye nomuthi. Umuthi, ingoma yahamba.
"Sibekwe kahle njengephayona ekuthuthukisweni nasekuqaliseni okusemthethweni kobuhlakani bokufakelwa, kusuka kokutholwa okuyisisekelo kwemvelo ukwenza ngcono ukuthuthukiswa kwezidakamizwa nokwenza izinqubo zokuthengwa kwemitholampilo zisebenze kahle ngokwezezimali. ukunakekela impilo. Indlela uhlelo lokunakekelwa kwezempilo olusethwa ngayo, "kusho yena.
Izipikha eziningana zigcizelele umqondo olula: Gxila kumsebenzisi (kulokhu, isiguli noma udokotela) nakho konke okunye kuzolandela. "Kubeka isiguli maphakathi nakho konke esikwenzayo," kusho uDkt Leh Lehmann, umqondisi wama-Bioethics esibhedlela sabesifazane. "Kudingeka sicabangele izidingo zabo nezinto eziza kuqala kubo."
Ukusuka kwesobunxele kuye kwesokudla: Izindaba ze-Anchor Mohana Ravightranath; UJessica Peter Lee ocwaningweni lweMicrosoft; USylvia Plevritis, uprofesa we-Biomedical Data Science, uxoxa ngeqhaza lobuhlakani bokufakelwa ocwaningweni lwezokwelapha. USteve Fish
Izikhulumi ephaneli, ezazihlanganisa uLehmann, eStanford University Medical Bioeheetheddred CO, MD, neSikhulu Esikhulu Lemitholampilo YaseMichael Howell, i-MD, saphawula ubunzima bezinhlelo zezibhedlela, egcizelela isidingo sokuqonda injongo yabo ngaphambi kokungenelela kwabo. Yisebenzise futhi uqinisekise ukuthi wonke ama-Systems athuthukisiwe ahlanganisa futhi alalele abantu abaklanyelwe ukusiza.
Ukhiye owodwa stud ukucaca: Kwenza kucace lapho imininingwane esetshenziselwa ukuqeqesha i-algorithm ivela, ukuthi injongo yokuqala ye-algorithm, nokuthi idatha yesikhathi esizayo izoqhubeka nokusiza i-algorithm ifunde, phakathi kwezinye izinto.
"Ukuzama ukubikezela izinkinga zokuziphatha ngaphambi kokuba zibe zimbi [izindlela] zithola indawo ephelele elula lapho wazi ngokwanele ngalobu buchwepheshe ukuzethemba okuthile kukho, kodwa hhayi ngaphambi [kwenkinga] kuyaqhubeka nokuyixazulula kungekudala." , Kusho uDenton Char. Umuntu ozongenela isayensi yezokwelapha, uprofesa ohambisana noMnyango Wezokwelapha Obuhle BasePediatric Anesteriology, umuthi osebenzisana naye kanye nemithi yezinhlungu. Enye isinyathelo esingukhiye, uthi, ukukhomba bonke ababambiqhaza abangathinteka ubuchwepheshe futhi banqume ukuthi bangathanda kanjani ukuphendula leyo mibuzo.
UJesse Ehrendeld, MD, uMongameli we-American Medical Association, uxoxa ngezinto ezine eziqhuba noma yiliphi ithuluzi lezempilo ledijithali, kubandakanya nalawo anikezwe amandla ubuhlakani bokufakelwa. Ngabe iyasebenza? Ngabe lokhu kuzosebenza esikhungweni sami? Ngubani okhokha? Ngubani obhekele?
UMichael Pfeffer, MD, Omkhulu Wemininingwane Omkhulu Wezokwazisa Omkhulu Wezempilo, waveza isibonelo sakamuva lapho izingqinamba eziningi zihlolwe khona phakathi kwabahlengikazi ezibhedlela zaseStanford. Odokotela basekelwa amamodeli amakhulu ezilimi ahlinzeka ngezichasiselo zokuqala zemiyalezo yeziguli engenayo. Yize le phrojekthi ingaphelele, odokotela abasize ekuthuthukiseni umbiko wezobuchwepheshe wokuthi imodeli incipha umthwalo wabo.
"Sihlala sigxile ezintweni ezintathu ezibalulekile: ukuphepha, ukusebenza kahle kanye nokufakwa. Singabantu odokotela. Sithatha isifungo sokungabi gciwane, "kusho uNina Vasan, uMd, uprofesa wezengqondo emtholampilo wesayensi yezengqondo nezokuziphatha, ojoyine uChar noPfeffer bajoyine iqembu. "Lokhu kufanele kube yindlela yokuqala yokuhlola lawa amathuluzi."
UNigam Shah, MBBS, Ph.D., uprofesa wezokwelapha kanye nesayensi yezokwelapha kanye nesayensi yesayensi, kwaqala ingxoxo ngesibalo esishaqisayo naphezu kwesexwayiso esifanelekile ezilalelini. "Ngikhuluma ngamanani ajwayelekile nezinombolo, futhi kwesinye isikhathi bavame ukuqondisa ngqo," kusho yena.
Ngokusho kukaShah, impumelelo ye-AI incike emandleni ethu okulibeka. "Ukwenza ucwaningo lwesayensi olufanele ngemodeli kuthatha cishe iminyaka eyi-10, futhi uma ngabe yilowo nalowo mhlangane kanye nobudlelwano bezinhlelo zobudlelwano kanye nokuhlalisana babefuna ukuvivinya futhi kungaba nzima kakhulu ukwenza isayensi efanele njengoba nathi sihlela isayensi efanele njengoba njengamanje sihlele Imizamo yethu kanye [nokuhlola]] Kubiza ama- $ 138 billion ukwenza isiqiniseko sokuthi wonke amasayithi ethu asebenza kahle, "kusho uShah. "Ngeke sikwazi ukukukhokhela lokhu. Ngakho-ke kudingeka sithole indlela yokwandisa, futhi kudingeka sandise futhi senze isayensi enhle. Amakhono amakhulu asendaweni eyodwa futhi amakhono okukhula akwenye, ngakho-ke sizodinga lolu hlobo lobambiswano. "
U-Askated Dean Yuan Ashley no-Littlered CO (reception) baya emhlanganweni wezempilo. Steve Fish
Ezinye izikhulumi e-Symentsium zithi lokhu kungatholwa kubambiswano lomphakathi-yangasese, njengokuthuthukiswa kwe-White House Executive entuthukweni ephephile, ephephile futhi ethembekile kanye nokusetshenziswa kwezobunhloli bokufakelwa kwezempilo kanye ne-CHAI). ).
"Ubambiswano lomphakathi oluzimele olunamandla amakhulu kunawo wonke phakathi kwe-academia, izinkampani ezizimele kanye nomkhakha womphakathi," kusho u-Laura Adams, umeluleki omkhulu weNational Academy of Medicine. Ubuye waphawula ukuthi uhulumeni angaqinisekisa ukwethenjwa komphakathi, kanye nezikhungo zezokwelapha zezokwelapha. Nikeza ubuchwepheshe obuqotho, nobuchwepheshe bezobuchwepheshe kanye nesikhathi sekhompyutha kunganikezwa ngumkhakha wangasese. "Sonke singcono kunanoma ubani kithi, futhi siyakubona lokho ... Ngeke sithandaze ukuthola amandla ka [intelligence okufakelwa] ngaphandle kokuthi siqonde ukuthi singaxhumana kanjani nomunye."
Izikhulumi eziningana zithi i-AI ibuye ibe nomthelela ocwaningweni, ukuthi ososayensi bayisebenzisa yini ukubheka imfundiso yemvelo, babikezela ukulandelana kwezinto eziphilayo, noma bazisize ukuba bafingqiwe noma babhale amaphepha esayensi.
"Leli yithuba lokubona okungaziwa," kusho uJessica Mega, i-MD, isazi senhliziyo eStanford University School of Medicine kanye nomsunguli wezinhlamvu zamagama. UMega ukhulume nge-hypersectral wataba, ethwebula izici zezithombe ezingabonakali esweni lomuntu. Umqondo ukusebenzisa ubuhlakani bokufakelwa ukuthola amaphethini kuma-pathology amaslayidi abantu abangakuboni lokho okukhombisa ukugula. "Ngigqugquzela abantu ukuba bamukele abangaziwa. Ngicabanga ukuthi wonke umuntu lapha uyakwazi umuntu onohlobo oluthile lwesimo sezokwelapha esidinga okuthile okungaphezu kwalokho esingakunikeza namuhla, "kusho uMejia.
AmaPanelists avumelane nalolo hlelo lwezobunhloli bokufakelwa luzohlinzeka ngezindlela ezintsha zokuhlonza nokulwa nezinqumo ezinohlonze, noma ngabe kwenziwa ngabantu noma ubuhlakani bokubonisa ukukhomba umthombo we-bias.
"Impilo ingaphezu kokunakekelwa kwezokwelapha nje," ama-panelist avumelana ngayo. Izipikha zigcizelele ukuthi abacwaningi bavame ukungakunaki izinqumo zezenhlalo zezempilo, njengesimo senhlalo yezenhlalo, ikhodi ye-zip, izinga lemfundo, kanye nobuhlanga, lapho beqoqa ababambiqhaza bezofunda. "I-AI isebenza ngempumelelo nje njengedatha lapho imodeli iqeqeshwa khona," kusho uMichelle Williams, uprofesa wezempi eHarvard University eHarvard University kanye noprofesa we-Associate of Epidemiology School. "Uma senza lokho esilwela ukukwenza. Thuthukisa imiphumela yezempilo futhi ususe ukungalingani, kufanele siqinisekise ukuthi siqoqa idatha esezingeni eliphakeme ekuziphatheni komuntu nasendaweni yezenhlalo nemvelo. "
INatalie Pageler, MD, uprofesa wezomtholampilo wezengane kanye nemithi, kusho imininingwane yomdlavuza ehlanganisiwe ivame ukungafaki idatha kwabesifazane abakhulelwe, yakha ukucwasana okungenakugwenywa kumamodeli kanye nokukhulisa izinto ezikhona ekunakekelweni kwezempilo.
UDkt David Magnus, uprofesa wezingane kanye nemithi, wathi kunanoma yibuphi ubuchwepheshe obusha, ubuhlakani bokufakelwa bungenza izinto zibe ngcono ngezindlela eziningi noma zizenze zibe zimbi kakhulu. Ingozi, kusho uMagnus, ukuthi izinqubo zezobunhloli zokufakelwa zizofunda ngemiphumela yezempilo engeyona eyayiqhutshwa izinqumo zezenhlalo zezempilo futhi ziqinisa leyo miphumela ngokuphuma kwazo. "Ubuhlakani bokufakelwa buyisibuko esibonisa umphakathi esiphila kuwo," kusho yena. "Ngiyethemba ukuthi ngaso sonke isikhathi lapho sinethuba lokukhanya ukukhanya engxabeni - ukubamba isibuko kuze kube yize sibe yisisusa sokuthuthukisa lesi simo."
Uma ungakwazanga ukuya e-Recal Workshop, ukuqoshwa kweseshini kungatholakala lapha.
IStanford University School of Medicine uhlelo oluhlanganisiwe lokunakekelwa kwempilo yezemfundo oluqukethe iStanford University School of Medicine kanye nezinhlelo zabantu abadala nezokunikezwa kwezempilo zokulethwa kwezingane. Ngokubambisana babona amandla aphelele e-biomedicine ngocwaningo lokubambisana, imfundo kanye nokunakekelwa kweziguli emtholampilo. Ngeminye imininingwane, vakashela i-Med.stanford.edu.
Imodeli entsha yezobunhloli yokufakelwa isiza odokotela nabahlengikazi esibhedlela saseStanford ndawonye ukwenza ngcono ukunakekelwa kweziguli.
Isikhathi Sokuposa: Jul-19-2024