Seite - 143 - in The Future of Software Quality Assurance

Responsible Software Engineering 143 risk of exclusion, and to situations which are characterized by asymmetries of power or information, such as between employers and workers, or between businesses and consumers. 3. Acknowledge that, while bringing substantial benefits to individuals and society, AI systemsalsoposecertainrisksandmayhaveanegative impact, includingimpactswhich maybedifficult toanticipate, identifyormeasure (e.g.ondemocracy, theruleof lawand distributive justice, or on the human mind itself.) Adopt adequate measures to mitigate these risks when appropriate, and proportionately to the magnitude of the risk.” Another initiative is AI4People: It is a multi-stakeholder forum that “brings togetherall stakeholders interested in shaping the societal impact ofAI—including the European Commission, the European Parliament, civil society organizations, industry and the media” [33]. The result is a living document with the following preamble: “We believe that, in order to create a Good AI Society, the ethical . . . should be embedded in the default practices of AI. In particular, AI should be designed and developed in ways that decrease inequality and further social empowerment, with respect for human autonomy, and increase benefits that are shared by all, equitably. It is especially important that AI be explicable, as explicability is a critical tool to build public trust in, and understanding of, the technology.” Theso-calledAlgo.Rules[34]defineanewapproachonhowtopromotesoftware trust systematically. It was developed by the think tank iRights.Lab together with several experts in the field. New rules define how an algorithm must be designed in order to be able to be evaluated with moral authority: above all, transparent, comprehensible in its effectsandcontrollable: 1. “Strengthen competency: The function and potential effects of an algorithmic system must be understood. 2. Defineresponsibilities:Anaturalor legalperson mustalwaysbeheldresponsible for the effects involved with the use ofan algorithmic system. 3. Document goals and anticipated impact: The objectives and expected impact of the use ofan algorithmic system must be documented and assessed prior to implementation. 4. Guarantee security: The security of an algorithmic system must be tested before and during its implementation. 5. Provide labelling:The use of an algorithmic system mustbe identified as such. 6. Ensure intelligibility:Thedecision-making processeswithinanalgorithmicsystemmust always be comprehensible. 7. Safeguard manageability: An algorithmic system must be manageable throughout the lifetimeof itsuse. 8. Monitor impact: The effects of an algorithmic system must be reviewed on a regular basis. 9. Establish complaint mechanisms: If an algorithmic system results in a questionable decisionoradecision thataffects an individual’s rights, itmustbepossible torequest an explanation and file a complaint.”