Trade-offs
Quality of services versus privacy:
Using personal data may improve public services by tailoring them based on personal characteristics or demographics, but compromise personal privacy because of high data demands.
Personalisation versus solidarity:
Increasing personalisation of services and information may bring economic and individual benefits, but risks creating or furthering divisions and undermining community solidarity.
Convenience versus dignity:
Increasing automation and quantification could make lives more convenient, but risks undermining those unquantifiable values and skills that constitute human dignity and individuality.
Privacy versus transparency:
The need to respect privacy or intellectual property may make it difficult to provide fully satisfying information about an algorithm or the data on which it was trained.
Accuracy versus explainability:
The most accurate algorithms may be based on complex methods (such as deep learning), the internal logic of which its developers or users do not fully understand.
Accuracy versus fairness:
An algorithm which is most accurate on average may systematically discriminate against a specific minority.
Satisfaction of preferences versus equality:
Automation and AI could invigorate industries and spearhead new technologies, but also exacerbate exclusion and poverty.
Efficiency versus safety and sustainability:
Pursuing technological progress as quickly as possible may not leave enough time to ensure that developments are safe, robust and reliable.
Reference:
WHITTLESTONE, Jess et al. Ethical and societal implications of algorithms, data, and artificial intelligence: a roadmap for research. London: Nuffield Foundation, 2019.