Human-centered AI Colloquium Steven S. Gouveia: Black-Boxes, AI Medicine, and Trust

The Human-centered Artificial Intelligence Colloquium Committee warmly invites you to attend the October colloquium on 4 October. The talk ‘Black-Boxes, AI Medicine, and Trust: An Ethical Exploration’ will be presented by Steven S. Gouveia (University of Minho). After the colloquium, you are invited to join us for drinks.

Artificial Intelligence in Medicine

The influence of Artificial Intelligence in Medicine is evolving rapidly in today’s society. The basic assumption behind this technology is to produce health practices more reliable, accurate, efficient, and cheaper than traditional medicine, based on human reasoning. The alternative pathway is to create machine-learning algorithms that will assist part or the totality of the processes of decision-making in medical contexts.

Although it is true that algorithms can increase the efficiency and reliability of many different medical processes, it is important to note that most of these technologies are based on complex and multifaceted kind of data and content. Because of that, most of these AI systems are pure ‘black-boxes’: the practitioner will be able to understand the inputs of the system and, then, the outputs. However, they will never have access to what happens ‘inside’ the system, turning the medical process an opaque process (epistemically) and careless (ethically).

Trust gap

Because of the nature of this technology, this new approach will create a ‘trust gap’ in the relationship between patients and medical experts: the nature of trust in that relationship will become doubtful and ambiguous.

The goal of this talk will be to analyse the main arguments in favour and against the use of ‘AI Medicine’, focusing on three ways of offering an answer to the problems raised by the use of ‘black-box’ artificial systems:

1. Restrict/constrain the development and use of AI Systems in Healthcare;
2. Accept the benefits of black-box AI Systems and ignore the consequences;
3. Create an ‘Explainable/Transparent’ Artificial Intelligence.

By analysing the positive and negative aspects of each of the three approaches that try to offer a cogent answer to the ‘trust gap’, we will be able to diagnose and identify the main relevant aspects that should be considered between care, on the one hand, and algorithmic process in healthcare, on the other hand.

Steven S. Gouveia

Steven S. Gouveia is a PhD (2021) from the University of Minho (Portugal). He was a visiting researcher in the Minds, Brain Imaging and Neuroethics Unit at Uni. Ottawa (2017 & 2019), where he was a PostDoc Research Fellow (2021-2022). He was a Researcher at UCP-Braga (2022-2023) and is currently leading a project (6 years) on the Ethics of Artificial Intelligence in Medicine at the MLAG of the Institute of Philosophy at the University of Porto, which has a partnership with Yale, Exeter and Helsinki.

Gouveia has published 14 academic books, the most recent being Thinking the New World: Conversations on AI. He hosted the documentary The Age of AI: The Documentary and has been a keynote speaker in countries such as South Korea, Argentina, Cyprus, among others. Since June 2023, he is an honorary professor from the Faculdad Medicina Andrés Bello (Chile), an award received alongside Laureate Nobel Sir Roger Penrose, and he is also a professor of online courses that feature some of the most important thinkers in the world, such Peter Singer, Noam Chomsky, or Anil Seth.