Send message to Author

Short abstract:

AI is transforming our lives. What can regulators do to ensure this technology is making the best for the users and society at large? The EU AI Act refers to specific risks, thereby risking of facing new risks. My aim is to offer a methodology for mapping AI risks to be addressed by regulators.

Long abstract:

The rise of AI has led many regulators around the world to realize that they need to intervene and safeguard their citizens from the potential harms of this technology. The “poster child” of the regulatory efforts is the EU’s AI Act, with the first draft published in April 2021. The Act takes a risk-based approach, listing specific AI applications and ranking them according to their potential harm. Fast forward to June 2023, when the European Parliament officially approved the Act. It turned out that the two-year gap between the draft and the approval necessitated some updates, most notably to challenges associated with LLM-based systems such as ChatGPT. The changes to the 2021 draft consumed intensive negotiations which only ended in December 2023. The risk is that new risks may emerge, and the regulation may become outdated even before it enters into force. Moreover, risks typically tend to surprise us when they materialize (Jasanoff, 2016). The ability to anticipate as many risks as possible is particularly important when dealing with regulation, which involves slow processes whose results should endure many years. This is the challenge to be addressed by this article. The aim is to offer a methodology for mapping AI risks to be addressed by future regulation. Postphenomenology can help us to map the risks by the relations associated with them, combined with new ethics and politics of technology (Verbeek 2011, Rosenberger 2017, Verbeek 2020, Boenink and Kudina 2020).