BSCC Focus: AI and Responsibility with Viveka Bonde
AI and Responsibility
As advances in AI occur at a high speed, many questions arise – not only of technical character but also economic, legal, ethical and societal issues arise, to mention a few. Because of this, as AI develops perhaps the most important need of AI advances is the need to think responsibility.
The panelists; Neelima Misra (Centigo), Anna Berggren (Trygghetsrådet), Anna Hjalmarsson (Artificial Solutions) and Minanda Fritz (LightNet Foundation), identified how important it is for an organization to prioritize, preferably throughout the whole AI value chain, values such as traceability, explainability, data governance, sustainability, accountability and ethics. Such values need to be addressed and identified both on a system level, for instance by ‘ethics by design’, as well as on an organizational level, to be able to create a good ground for the making of responsible business decisions.
To summarize some ‘hands on’ advice given by the panelists for the creation of a responsible decision within AI:
- Internal Communication; what organization values and goals are prioritized? The priorities and goals of the organization need to be communicated, to aid the employees and/or collaborators to make the right choices and to understand why such choices are being made and how.
- Provide guidance as how to evaluate AI; for instance, on a system level, how does the organization’s data scientists work with concepts such as bias and fairness, and what metrics are being used? On an organizational level, leaders need to build teams with different backgrounds and perspectives to avoid biases and promote fairness.
- Data governance – traceability in terms of the data being used; transparency as to data collection, data sources, data selection and training data. Any privacy concerns? Any ethical concerns? Legal and/or compliance concerns?
- Identifying the different kinds of malice that is feasible, not only in terms of the output, but throughout the whole AI value chain. For instance, any unethical means to achieve correct outcomes?
The success values suggested by the panelists for the creation of reliable business decisions within AI are traceability, explainability, data governance, sustainability, accountability and ethics. By abiding by such values, prosperous opportunities and benefits will be created for all people to enjoy such smart agency as AI, not only for the societal good, but also for the individual actor on the market.
Finally, the panelists all agreed that the most exciting AI applications are yet to be deployed!
Partner, Bonde Barzey Advokatbyrå