Isikhangiso

I-Generative Artificial Intelligence (AI): I-WHO ikhipha Isiqondiso esisha ekuphathweni kwama-LMM

WHO ikhiphe isiqondiso esisha mayelana nezimiso zokuziphatha kanye nokubusa kwamamodeli amakhulu e-multi-modal (ama-LMM) ukuze isetshenziswe ngendlela efanele ukuze kuthuthukiswe nokuvikela impilo yabantu. Ama-LMM wuhlobo lokukhiqiza olukhula ngokushesha ukuhlakanipha okungekhona okwangempela (AI) ubuchwepheshe obunezinhlelo ezinhlanu ezibanzi zezempilo in 

1. Ukuxilongwa nokunakekelwa kwezempilo, njengokuphendula imibuzo ebhaliwe yeziguli; 

2. Ukusetshenziswa okuqondiswa yisiguli, njengokuphenya izimpawu nokwelashwa; 

3. Imisebenzi yomabhalane neyokuphatha, efana nokubhala kanye nokufingqa ukuvakasha kweziguli ngaphakathi kwamarekhodi ezempilo ngogesi; 

4. Imfundo yezokwelapha nezobuhlengikazi, okuhlanganisa nokuhlinzeka abaqeqeshwayo ngokuhlangana kwesiguli okulingiswayo, kanye; 

5. Ucwaningo lwesayensi nokuthuthukiswa kwezidakamizwa, okuhlanganisa ukuhlonza izinhlanganisela ezintsha. 

Nokho, lezi zinhlelo zokusebenza ekunakekelweni kwezempilo zinezingozi zokukhiqiza izitatimende ezingamanga, ezinganembile, ezichemile, noma ezingaphelele, ezingalimaza abantu abasebenzisa ulwazi olunjalo ekwenzeni izinqumo zezempilo. Ngaphezu kwalokho, ama-LMM angase aqeqeshwe kudatha yekhwalithi engeyinhle noma echemayo, kungaba ngokohlanga, ubuzwe, amadlozi, ubulili, ubunikazi bobulili, noma ubudala. Kukhona futhi ubungozi obubanzi ezinhlelweni zezempilo, njengokufinyeleleka kanye nokufinyeleleka kwama-LMM asebenza kahle kakhulu. Ama-LMM angaphinde akhuthaze 'ukuchema okuzenzakalelayo' ngochwepheshe bezempilo neziguli, lapho amaphutha enganakwa abengabonakala noma ukukhetha okunzima kudluliselwe ngokungafanele ku-LMM. Ama-LMM, njengezinye izinhlobo ze AI, futhi basengozini yezingozi ze-cybersecurity ezingafaka engozini ulwazi lwesiguli noma ukwethembeka kwalawa ma-algorithms kanye nokuhlinzekwa kokunakekelwa kwezempilo ngokubanzi. 

Ngakho-ke, ukuze kwakhiwe ama-LMM aphephile nasebenzayo, i-WHO yenze izincomo kohulumeni nabathuthukisi bama-LMM. 

Ohulumeni banomthwalo wemfanelo oyinhloko wokumisa izindinganiso zokuthuthukiswa nokusatshalaliswa kwama-LMM, kanye nokuhlanganiswa kwawo nokusetshenziselwa izinjongo zempilo yomphakathi kanye nezokwelapha. Ohulumeni kufanele batshale imali noma bahlinzeke ngezingqalasizinda ezingenzi nzuzo noma zomphakathi, okuhlanganisa amandla ekhompuyutha namasethi edatha yomphakathi, okufinyeleleka konjiniyela abasemikhakheni yomphakathi, ezimele nengenzi-nzuzo, edinga abasebenzisi ukuthi bahambisane nezimiso zokuziphatha namagugu ukushintshanisa ukufinyelela. 

· Sebenzisa imithetho, izinqubomgomo kanye neziqondiso ukuqinisekisa ukuthi ama-LMM nezicelo ezisetshenziswa ekunakekelweni kwezempilo nemithi, kungakhathaliseki ubungozi noma inzuzo ehlobene AI ubuchwepheshe, ukuhlangabezana nezibopho zokuziphatha kanye nezindinganiso zamalungelo abantu ezithinta, isibonelo, isithunzi somuntu, ukuzimela noma ubumfihlo. 

· Yabela i-ejensi elawulayo ekhona noma entsha ukuthi ihlole futhi ivumele ama-LMM kanye nezicelo ezihloselwe ukusetshenziswa ekunakekelweni kwezempilo noma kwezokwelapha - njengoba izinsiza zivuma. 

· Ukwethula okuphoqelekile kocwaningo lwangemva kokukhishwa nokuhlolwa komthelela, okuhlanganisa ukuvikelwa kodatha namalungelo abantu, yizinkampani zangaphandle ezizimele lapho i-LMM isetshenziswa ngezinga elikhulu. Ukuhlolwa kwamabhuku kanye nokuhlolwa komthelela kufanele kushicilelwe 

futhi kufanele ifake imiphumela nemithelela ehlukaniswe uhlobo lomsebenzisi, okuhlanganisa isibonelo ngeminyaka, uhlanga noma ukukhubazeka. 

Ama-LMM aklanyelwe hhayi kuphela ososayensi nonjiniyela. Abasebenzisi okungenzeka kanye nabo bonke ababambiqhaza abaqondile nabangaqondile, okuhlanganisa abahlinzeki bezokwelapha, abacwaningi besayensi, ochwepheshe bezokunakekelwa kwezempilo neziguli, kufanele bahlanganyele kusukela ezigabeni zokuqala ze AI ukuthuthukiswa komklamo ohlelekile, obandakanyayo, osobala kanye namathuba anikezwayo okuphakamisa izindaba zokuziphatha, amazwi okukhathazayo kanye nokunikeza imibono AI isicelo esicutshungulwayo. 

Ama-LMM aklanyelwe ukwenza imisebenzi echazwe kahle ngokunemba okudingekayo nokwethembeka ukuze kuthuthukiswe amandla ezinhlelo zezempilo futhi kuthuthukiswe izithakazelo zesiguli. Onjiniyela kufanele futhi bakwazi ukubikezela nokuqonda imiphumela yesibili engaba khona. 

*** 

Source: 

I-WHO 2024. Izimiso zokuziphatha nokuphathwa kobuhlakani bokwenziwa kwezempilo: isiqondiso samamodeli amakhulu anezimo eziningi. Itholakala ngo https://iris.who.int/bitstream/handle/10665/375579/9789240084759-eng.pdf?sequence=1&isAllowed=y 

***

Ithimba le-SCIEU
Ithimba le-SCIEUhttps://www.ScientificEuropean.co.uk
I-Scientific European® | I-SCIEU.com | Intuthuko enkulu kwezesayensi. Umthelela esintwini. Izingqondo ezikhuthazayo.

Bhalisela iphephandaba lethu

Ukuvuselelwa ngazo zonke izindaba zakamuva, okunikezwayo kanye nezimemezelo ezikhethekile.

Imibhalo Edume Kakhulu

I-Molnupiravir iba Umuthi Womlomo Wokuqala Wokubulala Amagciwane ukufakwa kuZiqondiso eziphilayo ze-WHO...

I-WHO ibuyekeze imihlahlandlela yayo yokuphila mayelana ne-COVID-19 Therapeutics....

Umuthi Omusha Ovimbela Izimuncagazi Umalaleveva Ekutheleleni Omiyane

Sekuhlonziwe izithako ezingavimbela izinambuzane zikamalaleveva...

I-'Wireless Pacemaker'' Engenantambo Engahlonza futhi Ivimbele Ukubanjwa

Onjiniyela bakhe i-'brain pacemaker' engenantambo enga...
- Ukukhangisa -
94,232abalandeliLike
47,611AbalandeliLandela
1,772AbalandeliLandela
30SubscribersBhalisa