By Marcus Du Sautoy, The Belknap Press of Harvard University Press: 2019, 312 pp. paperback, $17.95. ISBN: 978-0-674-988132
The ubiquity of AI in contemporary life forces us to reckon with the transformative capabilities of technical systems that afford new information practices. Yet, to do it correctly is to not lose sight of both the deeply human and profoundly mathematical aspects of these assemblages. Marcus Du Sautoy’s latest offering, The Creativity Code: Art and Innovation in the Age of AI, succeeds in its attempt to explain from an analytical, mathematical perspective for the layperson, and provides readers with a personal-qua-scientific narrative about how the march of mathematics and marvel of thinking machines are closely interwoven with historical and socio-technical elements. This book is not overly technical and draws upon well-known cases in the history of mathematics and computing in order to further the reader’s understanding of both the subject and the innately creative contexts that paved the path for the ubiquity of contemporary AI.
Du Sautoy examines the classical algorithmic “imitation game” from which the Turing test derives, and cogently argues instead for the use of the Lovelace test, wherein computers are compelled towards creativity, as a more effective and strategic alternative. This position is developed with a history of algorithms using a frame of relationality that brings these human-machine assemblages to the fore. While Du Sautoy does explain how AI computation is superior to human cognition in the realm of games such as Go or Chess, he remains understandably skeptical about a computer’s capacity for categorically human-like creative independence.
Du Sautoy’s analysis takes a step back from the physical materiality of AI systems, and he instead directs his focus towards a human-machine relational frame in order to argue for what qualifies as creative intelligence. Chapter 4 introduces the “matching algorithm,” formulated by 1962 Nobel Prize winners David Gale and Lloyd Shapley, which set the stage for today’s widely used algorithmically sorted suggested content. However, Du Sautoy argues their Nobel-worthy puzzle still subsists from incomplete and inconsistent human input:
[M]odern algorithms that run our dating agencies are built on top of the puzzle that Gale and Shapley solved. The problem is more complex since information is incomplete. Preferences are movable and relative, and they shift from day to day. But essentially, the algorithms are trying to match people with preferences that will lead to stable and happy pairings. (57)
Here, Du Sautoy suggests that we implement relational aspects of decision making, which is at the heart of creative constructions and imaginative choices afforded in instances of incomplete, “movable and relative” information. In a broad stroke, he uses this platform to propose how qualified intelligence that does not simply choose but rather creates choices from relative preferencing can be categorized as a competitively embodied intelligence, whether organic or not.
Much of the core discussion in this book focuses on mathematical apologia that seeks to discern the relationship between art and mathematics. For Du Sautoy, the key juncture is ludic: both art and mathematics are, essentially, games to be played. Both must be played out in order for the result to be seen; while deductive reasoning dominates in social terms, the surprisingly creative nature of play and interplay ensures that a degree of abductive surprise will always be present in the act of creation. At the same time, Du Sautoy gives careful attention to the quotidian role of logos in society, as a way to logically prove that which is difficult to rhetorically argue. In the section “Origins of Proof,” he evokes Aristotelian and Euclidian experimentations of numerical axioms, where increasingly complex societies gamify forms of governance through the analogous treatment of numbers:
The beauty of the game is that, even though we are trying to capture how numbers and geometry work, we can view the whole game symbolically . . . . The Greeks were not content to say [the] interesting connection between odd numbers and square numbers had been observed so far without exception. They wanted logos to prove to them that it could never be otherwise—that it was a logical consequence of the basic axioms of governing how numbers work. (153-156)
Through this historical analogy, Du Sautoy contends that it is difficult to agree with an outcome produced by a logic one is unable to follow, but this may very well become the case as AI increases in capacity to creatively construct logical frameworks. In a rather future-forward discursive fashion, Du Sautoy’s discussion in this chapter structures intelligence through computational frameworks and encourages the reader to explore the independent agency in all things, beyond man and the rest of his carbon-based correlates.
What does it mean to be conscious? What does it mean to be creative? Scholars in their respective fields have struggled mightily to define these cavernous categories for humans and their experiences of the world, but what about machine intelligences? Now, having reached nothing near a consensus, we seek to do the same for technical systems that burgeon with relative agency. This text can be seen as a polite invitation for algorithms to come to the table, and if not to dine, then at least to engage in some light conversation. That said, Du Sautoy argues that we pay too much attention to the capacities of AI for such repartee; after all, until AI is creatively conscious (whatever that may mean), it remains a tool for extending human creativity.
This text by Du Sautoy is an unexpected primer to the history and nature of AI, offering for the otherwise uninitiated an introduction to the interweaving of algorithms and other forms of AI in our daily lives. It is a love letter to contemporary discussions of AI and its ancillaries within the history of mathematics and sociology, without contradiction. It is also a thoroughly persuasive book with a particularly humanistic approach to AI evangelism that is far-ranging but never strays far from its focus on the anthropic existence of AI.
—Jina Hong, University of California, Irvine