Patric Harting

The generative AI boom has led to many well-publicised worries, including over job losses, baked in discrimination, its use in crime, a flood of poor-quality content and misuse in journalism. But despite these concerns AI is here to stay, for the obvious reason that it is useful. In some cognitive domains AI tools have reached super-human performance and predictive power. This poses a number of difficult cognitive and ethical challenges, especially for conscientious professionals who do their best to uphold the highest ethical standards. In this paper I will look at one such issue and some of its implication. The issue arises when a typically reliable AI system makes predictions or suggests solutions that go beyond human understanding. How should professionals deal with such cases? Is it ethical to apply solutions one does not fully understand? Are there special considerations for time-critical or life-threatening situations? What, if anything, can professionals do to ameliorate their lack of understanding? How should professionals communicate about this issue, including to the public? Although this ethical challenge appears to be unavoidable, I argue that professionals can respond positively and proactively in several ways. 1) By looking to the virtues of honesty and epistemic humility. 2) Professionals can benefit from practical training in risk management 3) AI case studies should form a part of continuous professional development. 4) And professionals should engage with scientists and philosophers.