n the fringes of the internet, a 2010 thought-experiment took on a life of its own. Thanks in part to Elon Musk and other techno-futurists, it quickly jumped into the mainstream. Known as Roko’s Basilisk, it centered on the concept that an evil, godlike artificial intelligence, or AI, came into being in the future, but tormented those in the past that did not help bring about its creation. It relied on several assumptions, decision-making theories, and more, but according to some, even thinking about the Basilisk would doom you to an eternity of torture at the hands of this evil AI (it also assumed, to some degree, that we are all living in a simulation).
While very much a fringe thought-experiment, Roko’s Basilisk highlights the fears that many have about AI: that one day we will create our successors and in so doing, our doom. Yet, people happily talk with their Alexas, Siris, and other AI devices, content in believing that the worst thing that could happen is drunkenly ordering ramen noodle pool floats.
This is the challenge facing the general public when it comes to understanding AI and making AI policy. The lack of understanding of what AI is (and is not) fosters over-reliance on fear, hyperbole, and pop culture. T-Minus AI is a counter to this narrative, looking at artificial intelligence in an even-keeled fashion, something that is very much missing in today’s debate.
As author Michael Kanaan notes at the outset, this book is less about the science-fiction side of the house—the Terminators, Matrix, or Ex Machina—and more about the current state of play for artificial intelligence. The possibility of artificial general intelligence or superintelligence, to borrow from Nick Bostrom’s book title, is raised, but only in the context of just how far away humanity is from this singularity.
This is not to say that it isn’t a possibility or something worth discussing. Elon Musk would not be spending the number of resources he is, on his various AI-related endeavors, nor would Stephen Hawking have issued the warnings of a superintelligence were it not something worth considering. That breakout moment could occur, but for his book, Kanaan chooses to focus on the real-world, immediate implications.
Starting with the Big Bang
Divided into thirds, Kanaan’s book starts with the rise of human intelligence, covers the development—at a high level—of artificial intelligence, before finally turning to the geopolitical implications of artificial intelligence. The first third of the book establishes the evolution of human intelligence. At first glance, spending nearly 80 pages on the rise of human intelligence—starting from the Big Bang—seems out of place, until one appreciates that to understand artificial intelligence, one needs to understand what is intelligence and how it evolved. Indeed, as Carl Sagan said, quoted by Kanaan, “If you wish to make an apple pie from scratch, you must first invent the universe.” A rather puckish insight, but one that is not wrong, and one that informs Kanaan’s approach.
Building off of this foundation, Kanaan then explores the emergence of 21st century computing and the early days of artificial intelligence. He describes the role of gaming with AI development and recounts, in thrilling detail, the chess matches of Deep Blue versus Gary Kasparov, and AlphaGo’s dominance of the ancient game of territorial conquest and position. His exploration of how researchers are attempting to mimic the human brain clear and concise, and much more accessible for the lay-reader than Ray Kurzweil’s How to Create a Mind.
Beijing, Moscow, and AI
In the final third, Kanaan turns his attention to the international state of play on AI, returning to where he started, with Vladimir Putin’s oft-quoted soundbite on AI: “Whoever becomes the leader in this sphere [AI] will become the ruler of the world.” Kanaan’s overview of the implications of Russian and Chinese AI programs is refreshingly clear.
Kanaan’s analysis of China’s AI and machine learning-enabled social credit system is one of the most focused yet presented. Through near omnipresent digital surveillance, constant tracking and monitoring, and the cooperation of the BATs (Baidu, Alibaba, and Tencent) Beijing created the world of 1984 feared by George Orwell. Every behavior can be tracked, scored, and assessed leading to either incentives and rewards or disincentives and punishments.
Beijing even has an app for Xi Jinping thought that encourages daily use through in-app achievements and societal pressure from friends and colleagues via publication of scores. Were this system about controlling anti-social behavior, it would be bad enough, but it is enabling the wholesale persecution of the Uyghur population in Western China, cracking down on civil liberties in Hong Kong, and Beijing is keen to export the system to the West, finding customers in Ecuador and elsewhere.
China’s goal of digital and AI dominance is all the more alarming when one looks at their international campaign to have Huawei and ZTE build out 5G networks. If Beijing controls the pipes, the Chinese Communist Party will, potentially, have access to all of the data that flows through those networks. Extrapolate that one step further and one can see the potential national and economic security threat that control or even potential control represents. Huawei and ZTE may be cheap today, but the longer-term cost is far too high.
For Russia, domestic economic, industrial, and technological weaknesses—in Kanaan’s analysis—prevent it from being a top tier AI competitor. Indeed, when was the last time you purchased something, not intoxicating, that was “Made in Russia,” if ever? In the absence of that market-orientation, Moscow is working to deploy AI-enabled or augmented weapons systems, and using it to buttress their existing disinformation and propaganda efforts.
Interestingly, Kanaan doesn’t explore the cyber aspect of AI in his take on Russia’s programs. If anything, Russia has demonstrated a willingness to pursue its objectives through cyber warfare, probing the West’s infrastructure and signaling that, were it inclined to do so, it could wreak havoc. This undoubtedly will worsen with AI, machine learning, and automated offensive and defensive operations. Arguably, the cyber warfare aspects of AI are more concerning than “deep fakes”—artificially generated imagery—and propaganda, given the potential impact of such attacks.
An Unfinished Story
Reading T-Minus AI, the reviewer was reminded of a television program called Connections. First aired in 1978, but resurrected in the mid-90s (when the reviewer first saw it), Connections explored how innovations and inventions were linked, eschewing the traditional linear telling of progress and instead looking at progress as a series of interconnected (though disparate) events. The host, James Burke, was a delightful guide as the viewer explored the connections between polyethylene and Big Ben or a type of Dutch cargo ship and the invention of plastics.
Kanaan follows a similar, if more linear path, on the evolution of intelligence and artificial intelligence, but with the same enthusiasm and a similar wit to Burke. Artificial intelligence is here today, and is not some far off future, even if we don’t fully recognize or appreciate its presence.
Kanaan surveys the landscape of AI research and national programs for AI research. It is hard to find a country that isn’t pursuing some AI program as nearly all recognize its value, even if they don’t fully understand or appreciate what it is or will do. As he writes, AI is very much an unfinished story. How will we ensure democratic values in an AI society? How can we ensure that our human biases aren’t reflected in AI, as is already the case with algorithms reflecting their coders’ biases? How can we ensure that AIs don’t cater to our worst traits and behaviors? Indeed, we already see how algorithms from Facebook, YouTube, and others seem to drive individuals to extremism with alarming speed. One minute you are looking at fitness tips, the next you are on a QAnon page.
Where one wishes Kanaan dove deeper is on the socio-economic implications of artificial intelligence, automation, and machine learning. The potential economic dislocation of automation is no longer limited to low-skill, repetitive tasks. AI is being applied to law, medicine, and other high-skill jobs. This has the potential to upset the traditional model of employment to a degree never before seen in human history. P.W. Singer and August Cole in their fantastic book Burn-In (also reviewed for the Diplomatic Courier) explore a society in which this has happened, and it is alarming to say the least. For example, while it was already being discussed before COVID-19, universal basic income (UBI) is gaining some traction as a consequence of AI emergence, though it remains decidedly on the fringes of the political dialogue. Perhaps this is ground for a second, follow-on book.
Kanaan is a welcome guide to AI. He establishes a firm, but not overly complicated grounding in what AI is, how it has come about, and how it could shape the geopolitical future. T-Minus AI stands out from similar books for what it is not—it is not a breathless, hyperbolic recounting of humanity’s impending doom at the hands of a malicious machine gone rogue. This is very much a good thing. Rather, it is a cool-headed guide, like James Burke’s Connections, through the world of contemporary AI that isn’t burdened by overly complex programming languages, networking concepts, or philosophical treatises.
a global affairs media network
Demystifying the Ghost in the Machine
August 22, 2020
T-Minus AI | By Michael Kanaan | Benbella Books | August 2020.
O
n the fringes of the internet, a 2010 thought-experiment took on a life of its own. Thanks in part to Elon Musk and other techno-futurists, it quickly jumped into the mainstream. Known as Roko’s Basilisk, it centered on the concept that an evil, godlike artificial intelligence, or AI, came into being in the future, but tormented those in the past that did not help bring about its creation. It relied on several assumptions, decision-making theories, and more, but according to some, even thinking about the Basilisk would doom you to an eternity of torture at the hands of this evil AI (it also assumed, to some degree, that we are all living in a simulation).
While very much a fringe thought-experiment, Roko’s Basilisk highlights the fears that many have about AI: that one day we will create our successors and in so doing, our doom. Yet, people happily talk with their Alexas, Siris, and other AI devices, content in believing that the worst thing that could happen is drunkenly ordering ramen noodle pool floats.
This is the challenge facing the general public when it comes to understanding AI and making AI policy. The lack of understanding of what AI is (and is not) fosters over-reliance on fear, hyperbole, and pop culture. T-Minus AI is a counter to this narrative, looking at artificial intelligence in an even-keeled fashion, something that is very much missing in today’s debate.
As author Michael Kanaan notes at the outset, this book is less about the science-fiction side of the house—the Terminators, Matrix, or Ex Machina—and more about the current state of play for artificial intelligence. The possibility of artificial general intelligence or superintelligence, to borrow from Nick Bostrom’s book title, is raised, but only in the context of just how far away humanity is from this singularity.
This is not to say that it isn’t a possibility or something worth discussing. Elon Musk would not be spending the number of resources he is, on his various AI-related endeavors, nor would Stephen Hawking have issued the warnings of a superintelligence were it not something worth considering. That breakout moment could occur, but for his book, Kanaan chooses to focus on the real-world, immediate implications.
Starting with the Big Bang
Divided into thirds, Kanaan’s book starts with the rise of human intelligence, covers the development—at a high level—of artificial intelligence, before finally turning to the geopolitical implications of artificial intelligence. The first third of the book establishes the evolution of human intelligence. At first glance, spending nearly 80 pages on the rise of human intelligence—starting from the Big Bang—seems out of place, until one appreciates that to understand artificial intelligence, one needs to understand what is intelligence and how it evolved. Indeed, as Carl Sagan said, quoted by Kanaan, “If you wish to make an apple pie from scratch, you must first invent the universe.” A rather puckish insight, but one that is not wrong, and one that informs Kanaan’s approach.
Building off of this foundation, Kanaan then explores the emergence of 21st century computing and the early days of artificial intelligence. He describes the role of gaming with AI development and recounts, in thrilling detail, the chess matches of Deep Blue versus Gary Kasparov, and AlphaGo’s dominance of the ancient game of territorial conquest and position. His exploration of how researchers are attempting to mimic the human brain clear and concise, and much more accessible for the lay-reader than Ray Kurzweil’s How to Create a Mind.
Beijing, Moscow, and AI
In the final third, Kanaan turns his attention to the international state of play on AI, returning to where he started, with Vladimir Putin’s oft-quoted soundbite on AI: “Whoever becomes the leader in this sphere [AI] will become the ruler of the world.” Kanaan’s overview of the implications of Russian and Chinese AI programs is refreshingly clear.
Kanaan’s analysis of China’s AI and machine learning-enabled social credit system is one of the most focused yet presented. Through near omnipresent digital surveillance, constant tracking and monitoring, and the cooperation of the BATs (Baidu, Alibaba, and Tencent) Beijing created the world of 1984 feared by George Orwell. Every behavior can be tracked, scored, and assessed leading to either incentives and rewards or disincentives and punishments.
Beijing even has an app for Xi Jinping thought that encourages daily use through in-app achievements and societal pressure from friends and colleagues via publication of scores. Were this system about controlling anti-social behavior, it would be bad enough, but it is enabling the wholesale persecution of the Uyghur population in Western China, cracking down on civil liberties in Hong Kong, and Beijing is keen to export the system to the West, finding customers in Ecuador and elsewhere.
China’s goal of digital and AI dominance is all the more alarming when one looks at their international campaign to have Huawei and ZTE build out 5G networks. If Beijing controls the pipes, the Chinese Communist Party will, potentially, have access to all of the data that flows through those networks. Extrapolate that one step further and one can see the potential national and economic security threat that control or even potential control represents. Huawei and ZTE may be cheap today, but the longer-term cost is far too high.
For Russia, domestic economic, industrial, and technological weaknesses—in Kanaan’s analysis—prevent it from being a top tier AI competitor. Indeed, when was the last time you purchased something, not intoxicating, that was “Made in Russia,” if ever? In the absence of that market-orientation, Moscow is working to deploy AI-enabled or augmented weapons systems, and using it to buttress their existing disinformation and propaganda efforts.
Interestingly, Kanaan doesn’t explore the cyber aspect of AI in his take on Russia’s programs. If anything, Russia has demonstrated a willingness to pursue its objectives through cyber warfare, probing the West’s infrastructure and signaling that, were it inclined to do so, it could wreak havoc. This undoubtedly will worsen with AI, machine learning, and automated offensive and defensive operations. Arguably, the cyber warfare aspects of AI are more concerning than “deep fakes”—artificially generated imagery—and propaganda, given the potential impact of such attacks.
An Unfinished Story
Reading T-Minus AI, the reviewer was reminded of a television program called Connections. First aired in 1978, but resurrected in the mid-90s (when the reviewer first saw it), Connections explored how innovations and inventions were linked, eschewing the traditional linear telling of progress and instead looking at progress as a series of interconnected (though disparate) events. The host, James Burke, was a delightful guide as the viewer explored the connections between polyethylene and Big Ben or a type of Dutch cargo ship and the invention of plastics.
Kanaan follows a similar, if more linear path, on the evolution of intelligence and artificial intelligence, but with the same enthusiasm and a similar wit to Burke. Artificial intelligence is here today, and is not some far off future, even if we don’t fully recognize or appreciate its presence.
Kanaan surveys the landscape of AI research and national programs for AI research. It is hard to find a country that isn’t pursuing some AI program as nearly all recognize its value, even if they don’t fully understand or appreciate what it is or will do. As he writes, AI is very much an unfinished story. How will we ensure democratic values in an AI society? How can we ensure that our human biases aren’t reflected in AI, as is already the case with algorithms reflecting their coders’ biases? How can we ensure that AIs don’t cater to our worst traits and behaviors? Indeed, we already see how algorithms from Facebook, YouTube, and others seem to drive individuals to extremism with alarming speed. One minute you are looking at fitness tips, the next you are on a QAnon page.
Where one wishes Kanaan dove deeper is on the socio-economic implications of artificial intelligence, automation, and machine learning. The potential economic dislocation of automation is no longer limited to low-skill, repetitive tasks. AI is being applied to law, medicine, and other high-skill jobs. This has the potential to upset the traditional model of employment to a degree never before seen in human history. P.W. Singer and August Cole in their fantastic book Burn-In (also reviewed for the Diplomatic Courier) explore a society in which this has happened, and it is alarming to say the least. For example, while it was already being discussed before COVID-19, universal basic income (UBI) is gaining some traction as a consequence of AI emergence, though it remains decidedly on the fringes of the political dialogue. Perhaps this is ground for a second, follow-on book.
Kanaan is a welcome guide to AI. He establishes a firm, but not overly complicated grounding in what AI is, how it has come about, and how it could shape the geopolitical future. T-Minus AI stands out from similar books for what it is not—it is not a breathless, hyperbolic recounting of humanity’s impending doom at the hands of a malicious machine gone rogue. This is very much a good thing. Rather, it is a cool-headed guide, like James Burke’s Connections, through the world of contemporary AI that isn’t burdened by overly complex programming languages, networking concepts, or philosophical treatises.
