ISO-Stan­dard zur Ver­trau­ens­wür­dig­keit von AI-Systemen

Die hat einen Stan­dard (ISO/IEC TR 24028:2020) zur Ver­trau­ens­wür­dig­keit v0n AI-Syste­men ver­öf­fent­licht. Das Ziel des Doku­ments wird wie folgt umschrieben:

The goal of this docu­ment is to ana­ly­se the fac­tors that can impact the trust­wort­hi­ness of systems pro­vi­ding or using AI, cal­led here­af­ter arti­fi­cial intel­li­gence (AI) systems. The docu­ment brief­ly sur­veys the exi­sting approa­ches that can sup­port or impro­ve trust­wort­hi­ness in tech­ni­cal systems and dis­cus­ses their poten­ti­al app­li­ca­ti­on to AI systems. The docu­ment dis­cus­ses pos­si­ble approa­ches to miti­ga­ting AI system vul­nera­bi­li­ties that rela­te to trust­wort­hi­ness. The docu­ment also dis­cus­ses approa­ches to impro­ving the trust­wort­hi­ness of AI systems.

Kategorien Varia Tags ,