AiTech Agora

The ancient Greek word Agora refers to a public open space used for assemblies and markets. It captures the informal nature of our weekly meetings, a place where exchange of knowledge, ideas, and an engaging conversation takes place. The Agora is held on Wednesdays, beginning with lunch at 12:30, followed by a presentation by one of our team members or guest speakers and open discussion until 14:30.

If you wish to join the meetings and subscribe to the Agora mailing list, please contact us at aitech@tudelft.nl

Next meeting

January 29, 12:30 – 14:30 (RSVP before Monday noon)

Topic and speaker: A critical discussion on the book “Human Compatible – Artificial Intelligence and the Problem of Control”Luciano Cavalcante Siebert

Location: TPM, room A1.370 (Jaffalaan 5, 2628 BX Delft)

Abstract

The book “Human Compatible: Artificial Intelligence and the Problem of Control” explores the idea of intelligence in humans and in machines, describes the benefits we can expect, and outlines the AI breakthroughs that still have to happen before we reach superhuman AI. The distinguished AI researcher Stuart Russell (coauthor of the preeminent textbook “Artificial Intelligence: A Modern Approach”) spells out the ways humans are already finding to misuse AI, from lethal autonomous weapons to the manipulation of opinions on a global scale, and explains why we must ensure that we never lose control of machines more powerful than we are. He suggests that we can rebuild AI on a new foundation, according to which machines are designed to be inherently uncertain about the human preferences they are required to satisfy.

In this presentation, I will be examining and discussing the key concepts, assumptions, and propositions of Russel’s revised approach to AI from my personal point of view. Focusing on possible short-term ethical implications of (narrow) AI systems that may not be aligned with our moral values, I will discuss the problem of control over AI not only as a technical but as a socio-technical challenge, and also how the proposals in the book overlap/differ/relate to the concept of meaningful human control.