I-AI TRiSM: Ukuqinisekisa ukuthi i-AI ihlala ilungile

• About the Author:
• Igama lomlobi

  I-Quantumrun Foresight
• October 20, 2023

Isifinyezo sokuqonda

Ngo-2022, inkampani yocwaningo i-Gartner yethula i-AI TRiSM, emele i-AI Trust, Risk, and Security Management, ukuze kuqinisekiswe ukubusa nokuthembeka kwamamodeli e-AI. Uhlaka luqukethe izinsika ezinhlanu: ukuchazeka, ukusebenza kwemodeli, ukutholwa kokudidayo kwedatha, ukumelana nokuhlaselwa kwezitha, nokuvikelwa kwedatha. Umbiko ugqamisa ukuthi ukuphathwa kabi kwezingozi ze-AI kungaholela ekulahlekelweni okukhulu kanye nokwephulwa kwezokuphepha. Ukusebenzisa i-AI TRiSM kudinga ithimba elisebenza ngokubambisana elivela kwezomthetho, ukuthobela, i-IT, nokuhlaziya idatha. Lolu hlaka luhlose ukwakha isiko elithi "I-AI Enesibopho," eligxile ekukhathazekeni kwezimiso zokuziphatha nezomthetho, futhi kungenzeka libe nomthelela ekuthambekeni kokuqashwa, imithetho kahulumeni, nokucatshangelwa kokuziphatha ku-AI.

Umongo we-AI TRiSM

Ngokuka-Gartner, kunezinsika ezinhlanu ze-AI TriSM: ukuchazeka, Ukusebenza Kwemodeli (ModelOps), ukutholwa kwedatha ngokudidayo, ukumelana nokuhlaselwa kwezitha, nokuvikelwa kwedatha. Ngokusekelwe ekuqageleni kukaGartner, izinhlangano ezisebenzisa lezi zinsika zizobona ukukhuphuka okungamaphesenti angama-50 ekusebenzeni kwazo kwemodeli ye-AI ngokuphathelene nokutholwa, izinjongo zebhizinisi, nokwamukelwa kwabasebenzisi ngo-2026. Ukwengeza, imishini enamandla e-AI izokwakha amaphesenti angu-20 abasebenzi bomhlaba. futhi sinikele ngamaphesenti angama-40 omkhiqizo wezomnotho usuwonke ngo-2028.

Okutholwe ocwaningweni lukaGartner kuphakamisa ukuthi izinhlangano eziningi zenze amakhulu noma izinkulungwane zamamodeli e-AI abaphathi be-IT abangakwazi ukuwaqonda noma ukuwahumusha. Izinhlangano ezingalawuli ngokwanele ubungozi obuhlobene ne-AI zivame kakhulu ekuhlangabezaneni nemiphumela engathandeki kanye nokuphulwa kwemithetho. Amamodeli angahle angasebenzi njengoba kuhlosiwe, okuholela ekwephulweni kwezokuphepha nobumfihlo, kanye nokulimala kwezezimali, komuntu ngamunye, kanye nesithunzi. Ukuqaliswa okungalungile kwe-AI kungase futhi kubangele izinhlangano zenze izinqumo zebhizinisi ezingafanele.

Ukuze kusetshenziswe ngempumelelo i-AI TRiSM, ithimba elisebenza ngokubambisana kwezomthetho, ukuthobela, ukuphepha, i-IT, nabasebenzi bokuhlaziya idatha liyadingeka. Ukusungula ithimba elizinikele noma ithimba elisebenzayo elinokumelwa okufanele endaweni ngayinye yebhizinisi ebandakanyeka kuphrojekthi ye-AI nakho kuzoletha imiphumela emihle. Kubalulekile futhi ukuqinisekisa ukuthi ilungu ngalinye leqembu liyiqonda ngokucacile indima yalo nezibopho zalo, kanye nezinjongo nezinjongo zohlelo lwe-AI TRiSM.

Umthelela ophazamisayo

Ukwenza i-AI iphephe, uGartner uncoma izinyathelo ezimbalwa ezibalulekile. Okokuqala, izinhlangano zidinga ukubamba ubungozi obungaba khona obuhlobene ne-AI nokuthi zingabunciphisa kanjani. Lo mzamo udinga ukuhlolwa okubanzi kwengozi okungabheki nje kuphela ubuchwepheshe uqobo kodwa nomthelela wabo kubantu, izinqubo, kanye nemvelo.

Okwesibili, izinhlangano zidinga ukutshala imali ekubusweni kwe-AI, okuhlanganisa izinqubomgomo, izinqubo, nokulawula ukulawula ubungozi be-AI. Leli su lihlanganisa ukuqinisekisa ukuthi izinhlelo ze-AI zisobala, ziyachazeka, ziyaziphendulela, futhi ziyahambisana nemithetho neziqondiso ezifanele. Ukwengeza, ukuqapha okuqhubekayo nokuhlolwa kwamamodeli we-AI kubalulekile ukuze kukhonjwe futhi kuncishiswe noma yiziphi izingcuphe ezingase zivele ngokuhamba kwesikhathi. Okokugcina, izinhlangano zidinga ukuthuthukisa isiko lokuphepha kwe-AI, ukukhuthaza ukuqwashisa, imfundo, nokuqeqeshwa phakathi kwabasebenzi nababambiqhaza. Lezi zinyathelo zihlanganisa ukuqeqeshwa ngokusetshenziswa kokuziphatha kwe-AI, ubungozi obuhlobene ne-AI, kanye nendlela yokuhlonza nokubika ngayo izinkinga noma ukukhathazeka. 

Le mizamo izoholela ekutheni izinkampani eziningi zakhe iminyango yazo ye-AI enesibopho. Lolu hlaka lokuphatha oluvelayo lubhekana nezithiyo zomthetho nezokuziphatha ezihlobene ne-AI ngokubhala phansi ukuthi izinhlangano zibhekana kanjani nazo. Uhlaka kanye nezinhlelo ezihambisana nalo zifuna ukuqeda ukungaqondakali ukuze kuvinjelwe imiphumela emibi ebingahlosiwe. Izimiso zohlaka olunesibopho se-AI zigxile ekuklameni, ekuthuthukiseni, nasekusebenziseni i-AI ngezindlela ezizuzisa abasebenzi, ezinikeza inani kumakhasimende, futhi zibe nomthelela omuhle emphakathini.

Imithelela ye-AI TRiSM

Imithelela ebanzi ye-AI TRiSM ingafaka: 

• Njengoba i-AI TRiSM iba ngobaluleke kakhulu, izinkampani zizodinga ukuqasha abasebenzi abanamakhono amaningi abanolwazi kulo mkhakha, njengabahlaziyi bezokuphepha be-AI, abaphathi bezingcuphe, kanye nezazi zokuziphatha.
• Ukucatshangelwa okusha kwezimiso zokuziphatha nokuziphatha, njengesidingo sokubeka izinto obala, ukulunga, nokuziphendulela ekusebenziseni amasistimu e-AI.
• Ukuqanjwa okuthuthukisiwe kwe-AI okuvikelekile, okuthembekile, futhi okuthembekile.
• Ingcindezi eyengeziwe yokulawula kukahulumeni ukuze kuvikelwe abantu nezinhlangano ezingozini ezihlobene nezinhlelo ze-AI.
• Ukugxila kakhulu ekuqinisekiseni ukuthi izinhlelo ze-AI azichemi kumaqembu athile noma abantu abathile.
• Amathuba amasha alabo abanamakhono e-AI futhi okungenzeka kuxoshwe labo abangenawo.
• Ukwenyuka kokusetshenziswa kwamandla kanye nomthamo wokugcina idatha wedatha yokuqeqeshwa ebuyekezwa njalo.
• Izinkampani eziningi ziyahlawuliswa ngokungalandeli izindinganiso ze-AI Zokuzibophezela emhlabeni jikelele.

Imibuzo okufanele icatshangelwe

• Uma usebenza ku-AI, inkampani yakho iwaqeqesha kanjani ama-algorithms ayo ukuthi abe nesimilo?
• Yiziphi izinselelo zokwakha izinhlelo ze-AI eziPhethe?