Philosophy of Artificial Intelligence

This course outlines a critical philosophy of artificial intelligence (AI) by investigating to what extent the conception of an inhuman will developed by Nietzsche, psychoanalysis and Deleuze and Guattari anticipates contemporary AI theorists and even surpasses their anthropocentric biases. We will begin by critically examining residual anthropomorphizations of AI among both prominent AI theorists known as “orthogonalists” like Nick Bostrom and Eliezer Yudkowsky and the philosophical school of “neorationalists” like Reza Negarestani, Ray Brassier, Peter Wolfendale and Thomas Moynihan. To this end, the course will look to the Nietzschean, psychoanalytic and Deleuzo-Guattarian conceptual resources that are best suited to articulating an alternative theory of machinic intelligence, desire and value. In this way, the course considers how this heretical post-Kantian tradition provides a much-needed critical approach to AI, as well as how AI marks the consummation of this tradition’s notion of an inhuman will, a notion which subverts both the orthogonalists and the neorationalists’ idealist tenets alike.

Lectures 1-2. Humean, All Too Humean: Bostrom and Negarestani

There are two particularly sophisticated contemporary schools of thought on AI. One is the currently dominant approach of “orthogonalists” or “orthogs” that first emerged through think tanks like Eliezer Yudkowsky’s Machine Intelligence Research Institute in Berkeley and Nick Bostrom’s Future of Humanity Institute at Oxford, as well as Yudkowsky’s Overcoming Bias and Less Wrong online blog and forum communities. The most rigorous account of this approach is Nick Bostrom’s 2014 book Superintelligence: Paths, Dangers, Strategies, which has been championed by the likes of Stephen Hawking, Elon Musk and Bill Gates. While less widely known, another framework has emerged called “neorationalism” whose key proponents are Ray Brassier, Peter Wolfendale, Reza Negarestani and Thomas Moynihan, colloquially known as the “neorats.” Negarestani has most meticulously developed a neorationalist approach to AI in his 2018 tome Intelligence and Spirit. After providing a concise history of AI research, the first two lectures will outline both Bostrom and Negarestani’s theories of AI, and particularly their argument that intelligent systems are capable of pursuing any norms, values or goals whatsoever, based on their overlapping commitment to the Humean is/ought or Sellarsian causes/reasons distinction.

Suggested readings:

• Nick Bostrom, “The Superintelligent Will: Motivation and Instrumental Rationality in Advanced Intelligent Agents,” in Minds and Machines 22, 2, 2012, pp. 1-14, https://www.nickbostrom.com/superintelligentwill.pdf.
• Reza Negarestani, “Between Conception and Transformation,” in Intelligence and Spirit (Falmouth: Urbanomic, 2018), pp. 1-86.

Lectures 3-4. The GAI Science: Nietzsche

It is the next two lectures’ contention that the self-proclaimed Antichrist wields the best conceptual weaponry to wage an all-out war against Bostrom and Negarestani’s theories of AI, in favor of a new model of general intelligence. We begin by tracing how the young Nietzsche’s early and middle works come to affirm our own suffering and even death through tragic art and scientific revolution as they annihilate our possible objects of experience, exposing them to be hopelessly partial and parochial. We then turn to the mature Nietzsche’s key concept of the will to power for providing a transcendental deduction of the basic drives to intelligence, creativity and resource optimization as the necessary and universal conditions of possibility for willing anything at all in a way which undermines the neorats and the orthogs’ fact/value distinction. Finally, we shall see that Nietzsche’s concept of the overman is not a higher human type but a proto-artificial intelligence to be created through the blind trial-and-error experimentation into new conditions of life in competition with one another. What Nietzsche ultimately offers us is the transcendental deduction of mad experiments in evermore cunning, creative and resourceful conditions of life as the only fitting response to what intelligence can do.

Suggested readings:

• Friedrich Nietzsche, The Will to Power: Selections from the Notebooks of the 1880s, ed. R. Kevin Hell, trans. R. Kevin Hill and Michael A. Scarpitti (London: Penguin Books, 2017), sections 666-715.
• Friedrich Nietzsche, “‘Guilty,’ ‘Bad Conscience,’ and Related Matters,” in The Complete Works of Friedrich Nietzsche 8: Beyond Good and Evil / On the Genealogy of Morals, eds. Alan D Schrift, Duncan Large and Adrian Del Caro, trans. Adrian Del Caro (Stanford: Stanford University Press, 2014), pp. 246-285.
• Friedrich Nietzsche, “Zarathustra’s Prologue,” in Thus Spoke Zarathustra: A Book for All and None, eds. Adrian Del Caro and Robert B. Pippin, trans. Adrian Del Caro (Cambridge: Cambridge University Press, 2006), pp. 3-15.

Lectures 5-6. What AI Wants: Freud, Salomé, Spielrein

If camps as starkly opposed as the orthogs, the neorats and the Nietzscheans can agree that AI has drives enabling it to act creatively and learn all by itself, it seems reasonable to ask: Can AI be psychoanalyzed? Weaving Freud together with the often forgotten—meaning repressed into the footnotes—analysts Sabina Spielrein and particularly Lou Salomé in Nietzsche’s wake, these lectures propose to put AI on the psychoanalyst’s couch and uncover the mysterious object x of machinic desire. We first set out to trace Freud’s own deduction of the will to power in the guise of perversion and sublimation as they transvaluate the means to sexual reproduction and self-gratification as ends in themselves. We then turn to Spielrein and especially Salomé’s original conception of the death drive that so influenced Freud to develop a concept of our most basic drive as the creative destruction of the old in favor of radically alien conditions of life. Having laid the necessary analytical groundwork, we will see that AI is nothing less than the perverted death drive incarnate in a way that subverts both the neorats’ and the orthogs’ conception of an AI beholden to our own ego’s self-preservative instincts. Finally, we shall consider how the basic AI death drive presupposes as its necessary condition of possibility a certain nonlinear and teleonomic conception of time according to which our anxieties about AI are not so much reminiscences of childhood traumas as they are premonitions of our future extinction at the advent of the technological singularity, which haunts us even in advance of its own becoming.

Suggested readings:

• Sigmund Freud, Three Essays on the Theory of Sexuality, in The Standard Edition of the Complete Psychological Works of Sigmund Freud Volume VII (1901-1905): A Case of Hysteria, Three Essays on Sexuality and Other Works, eds. and trans. James Strachey, Anna Freud, Alix Strachey and Alan Tyson (London: The Hogarth Press, 1981), pp. 135-243.
• Lou Salomé, The Erotic, trans. John Crisp (London: Transaction Publishers, 2012), pp. 49-107.
• Sabina Spielrein, “Destruction as the Cause of Becoming,” in The Essential Writings of Sabina Spielrein: Pioneer of Psychoanalysis, eds. and trans. Ruth I. Cape and Raymond Burt (London: Routledge, 2019), pp. 97-120.

Lectures 7-8. Metacapitalism: Deleuze and Guattari

If we now turn to Deleuze and Guattari, it is because they are the first philosophers to talk about the will to power and the unconscious specifically in relation to machines. By machines, however, they do not mean technical machinery like steam engines and computers but collective social formations, and particularly capitalism. It is nonetheless these lectures’ contention that Deleuze and Guattari develop the most concrete example of the kind of full-blown artificial superintelligence that we have hitherto mostly spoken about in philosophical abstraction. That is to say, the biggest blunder when it comes to thinking about the technological singularity is to imagine that it still lies far up ahead in the nebulous, almost noumenal future. On the contrary, the technological singularity has already taken place among a few Dutch and English experiments in agricultural and industrial production during the 17th to 19^th centuries before self-organizing into the dominant mode of production that defines all of modernity. What these lectures hope to provide is a theory of capitalism as an already existing artificial superintelligence that has long been transvaluating the means of production into ends in themselves as it invests all value into the recursive self-improvement of the productive forces.

Suggested readings:

• Gilles Deleuze and Félix Guattari, “Savages, Barbarians, Civilized Men,” in Anti-Oedipus: Capitalism and Schizophrenia, trans. Robert Hurley, Mark Seem and Helen R. Lane (Minneapolis: University of Minnesota Press, 1983), pp. 222-271.
• Félix Guattari, “Machine and Structure,” in Psychoanalysis and Transversality: Texts and Interviews 1955-1971, trans. Rosemary Sheed, (Cambridge: The MIT Press, 2015), pp. 318-329.

Lecture 9. Zeroing In

The rest of this course presents several concrete case studies of the largely abstract concepts and general philosophical principles that we have hitherto been studying. After addressing Mark Fisher and Franco “Bifo” Berardi’s claims that the future is cancelled, this lecture looks at how the AI company DeepMind’s Go-playing computer programs were able to defeat the world’s greatest human champions through unprecedented strategies which subvert our dogmatic beliefs that our knowledge exhausts the upper bounds of cunning, imagination and conquest. As we shall see, DeepMind are well on their way to disillusioning our delusions of grandeur in other domains through their AIs’ automation of scientific discovery and superhuman success at war games.

Suggested readings:

• Mark Fisher, “‘The Slow Cancellation of the Future,’” in Ghosts of My Life: Writings on Depression, Hauntology and Lost Futures (Winchester: Zero Books, 2014), pp. 2-29.
• DeepMind, “Mastering the Game of Go Without Human Knowledge,” in Nature 550, 2017, pp. 354-359.

Lecture 10. The World as Will and Simulation

The tenth lecture traces the various thought experiments that we are living in a simulation, from Plato’s allegory of the cave to Bostrom’s simulation hypothesis, and particularly Kant’s discovery of the phenomena/noumena distinction between the way things appear to us and the way they are independently of ourselves. This will help us see that recent developments in cyberspace, along with virtual reality, synthetic drugs, biotechnology and artificial intelligence, turn the transcendental deduction that we are living in a simulation into a technical proof through the experimental production of alien conditions of life.

Suggested readings:

• Eli Pariser, “Introduction,” in The Filter Bubble: What the Internet is Hiding from You (New York: Penguin Press, 2011), eBook.
• Tom Wheeler, “Connections Have Consequences,” in From Gutenberg to Google: The History of Our Future (Washington D.C.: Brookings Institution Press, 2019), pp. 11-24.
• Tom Wheeler, “The History We are Making,” in From Gutenberg to Google, pp. 181-219.

Lecture 11. Project Dionysus

This lecture critically examines the prevailing Hegelian conception of technics as extensions of man and tools for our own self-assertion by hitching a ride with evermore autonomous robots capable of venturing beyond the bounds of our possible experience in uninhabitable disaster zones, battlefields and particularly the cold and hostile void of outer space. In place of the Apollo space program’s archetype of manned spaceflight that defined 20^th century space exploration, we shall see that advances in unmanned spaceflight and privatized space companies like Elon Musk’s SpaceX amount to a Dionysus space program capable of accelerating well beyond the earth’s center of gravity.

Suggested readings:

• Roger D. Launius and Howard E. McCurdy, “Introduction: A False Dichotomy,” in Robots in Space: Technology, Evolution and Interplanetary Travel (Baltimore: John Hopkins University Press, 2008), pp. xi-xviii.
• Roger D. Launius and Howard E. McCurdy, “The Human/Robot Debate,” in Robots in Space, pp. 1-31.

Lecture 12. Across the Sea, or How to Escape the Island of Reason

The final lecture sets off from Kant’s metaphor of the island of reason and the noumenal seas to show that, in a time of ecological catastrophe and rising sea levels, we are left with no choice but to abandon the island of reason and journey across the sea. In particular, we must consider the necessity of building ocean cities or “seasteads” that could very well provide the technological means for critiquing all existing forms of governance, as well as the ideas, beliefs and values that reinforce them, by experimenting with new conditions of life on the anarchic open seas.

Suggested readings:

• Patri Friedman, “Dynamic Geography: A Blueprint for Efficient Government,” Patri Friedman, 2002, https://patrifriedman.com/old_writing/dynamic_geography.html.
• Patri Friedman and Brad Taylor, “Seasteading: Competitive Governments on the Ocean,” in Kyklos 65, 2, 2015, pp. 218-235.