Lenartowicz, M. et al. (2021). What is Intelligence? In: M. Lenartowicz & W.D.R. Weinbaum (Eds.), The Practice of Thinking: Cultivating the Extraordinary. Academia Press (forthcoming) What is Intelligence? Marta Lenartowicz, Weaver D.R. Weinbaum, Francis Heylighen, Lotte Van Lith, and Veerle Meurs A ‘Thinking Studio’ seminar Marta Lenartowicz: The ambition of our initiative is to mobilise a cognitive power which we are referring to as 'extraordinary intelligence'. As many of you — participants of the project — have been pointing out, this ambition tends to evoke considerable attraction and equally considerable resentment, or at least reservation. The objections are partly ethical and political, partly cultural and psychological. When critiqued conceptually, the phrasing also calls for a serious re-visiting of the notion of intelligence. The 'extraordinary' we are after turns out to be less problematic: the way we put it, it is a progressive, dynamic term. 'Extraordinary' does not come to mean 'greater than', along a pre-established scale of measurement or relative to a baseline in a population, but rather 'greater again, and yet again', displaying qualities that outgrow themselves. The extraordinary intelligence would then be a capability that continuously exceeds its own limits, proving to be more insightful, more farsighted, and more potent, than one might normally project. But what is 'intelligence'? Clearly, we cannot be referring here to what is captured by IQ tests. The Intelligence Quotient is an iconic psychometric construct and the primary feature of such measurement methods is their reliable reference to traits that are persistent — ideally throughout the lifetime of the individuals assessed. If our interest is in an ever-changing, ever-growing intelligence, we cannot be applying a measure that by its very definition seeks to trace an invariant. The whole idea would become an oxymoron. But, if it is not that invariant predictable aspect of intelligence, what is it, then, that are we speaking about? Weaver D.R. Weinbaum: The so called invariant expression of intelligence is not necessarily intelligence in the deeper, more profound, sense of the word, as it is always the case that the former externalises the latter. In order to appreciate this, it is enough to venture out from the context of psychometrics, where intelligence is benchmarked, to the context of Artificial Intelligence where it is attempted to be produced. Even in the context of human intelligence, the range of definitions and understandings is much broader than it is conventionally assumed. There is a long debate on the nature of intelligence or even about whether it is a single concept. Shane Legg and Marcus Hutter (2007) have compiled a handy collection of eighteen general, thirty five psychological and eighteen AI-research based different definitions of intelligence, and many of them are not presuming that what they are referring to is invariant. Marta Lenartowicz: That is a lot of definitions. What are the most general formulations? Francis Heylighen: Generally speaking, intelligence characterises the mental ability of some agent. 'Mental' means not physical: for example not the ability to move things, but the ability to decide what you would be moving. 'Agent' here is meant very broadly: it can be a human individual, or more generally an organism, but it can also be a group or an organisation of people, an AI program, a robot, or some combination of those. An agent is a system that makes decisions in order to act in the real world. The most common definition of intelligence is that it is the ability