The Master Algorithm by Pedro Domingos

The Master Algorithm: How the Quest for the Ultimate Learning Machine Will Remake Our World by Pedro Domingos (Basic Books, 2015) 329 pages

TMA Domingos

When was the last time you saw a 12-word subtitle, I ask you? But after reading the book, I think a better subtitle would be “How Professor Domingos Took Bayes’ Theorem and Generalized It for Big Data.” And that’s still 12 words.

This is the “#1 Bestseller” in “Mathematical & Statistical Software” at Amazon. Which makes me sort of wonder how many categories they actually have and what the, say, #10 bestseller in that category might be.

Let me be clear: if you want to understand current research into artificial intelligence (AI), read this book. You might not understand everything, especially the discussion of Bayes theorem, but you’ll know enough to cut through the misconceptions.

What misconceptions? The ones about AI that I keep seeing online and hear when I talk to other people. Full disclosure: I have done PhD-level work in AI and work for a high-tech Silicon Valley company. I used to be a complete AI skeptic, but Domingos has convinced me not to be and that the whole field would be better off if it were called “machine learning” instead of AI, for reasons I will detail later.

What kind of AI misconceptions? Well, I saw one today from Tech Crunch (who should know better). In an article on “programming hate into AI” (HERE), the author says things like anything below the level of “coding in empathy” into an AI agent results in a “follower machine.”

The fallacy in this way of thinking is that you don’t really program anything “into” an AI agent. The whole point is to create a learning environment through rules or analogies or other methods and then “train” it to compile knowledge that can be used to answer questions about the world. For example, if we want an AI agent to be an expert in fruit, we can make rules like “if it’s round and red and…, then it’s an apple” or associations like “an apple has the characteristics round and red and…” or some other form of learning process. Then we can show the AI a picture and ask “Is this thing an apple?” and, if it’s done right, the response would be something like “it sure looks like an apple to me.” Intelligent robots and AI agents, as the book’s blurb mentions, “program themselves.”

And that’s the value of this book: it lets you cut through the nonsense written today about the future of AI and figure out what is really going on in the field today. Domingos is a professor and, like any good professor, he does a terrific job of introducing and explaining complex concepts to people who are not familiar with them.

The book surveys five different approaches to machine learning AI and attempts to combine them into a single “Master Algorithm” based on Domingos’s work with Bayes theorem and Markov logic networks as a unifier of all previous methods to achieve AI (page 246). Here are the five areas that Domingos details (they are nicely related on page 54, but the descriptions I took from page 291):

Symbolists who deal in “inverse deduction.” Starting on page 57, Domingos goes through the types of rules of induction that make up spam filters and the algorithms that banks use to detect fraud (“If your credit card was used to purchase one dollar of gas, it was stolen”, page 69). Is it 100% certain? No, and the problem over overfitting is a common one in AI (page 71). But the more apples our AI sees, the more it will be able to eliminate other round, red fruits like cherries or strawberries.

Connectionists who deal in “backpropagation.” Starting on page 93, Domingos goes over researchers who create AI algorithms based on “neural networks” and how the brain functions. One key concept that contributes to Domingos’s approach here is the “S-curve” (page 106). So our AI apple-detector will often err at first, then get a lot better, and then improve hardly at all.

Evolutionaries who deal in “genetic algorithms.” Starting on page 121, Domingos explores how learning systems can evolve themselves to become better. These are the people who create robot warriors that make versions of themselves that are smarter and smarter using 3D printers, a project that the military actually considered, according to Domingos. Domingos uses this as the starting point for his “Master Algorithm.”

Bayesians who deal in “probabilistic inference.” Starting on page 143, Domingos gets to the heart of his method, which is based on Markov logic networks. He derives MLNs from Bayes’s method of refining and weighing evidence (page 171). I have never seen a better explanation of what Bayes’ theorem actually does in action. Basically, the more real apples our apple AI sees, as opposed to plastic models or drawings, the better it should get (but it can never be 100% correct, because Bayes applies).

Analogizers who deal in “support vector machines.” Starting on page 177, Domingos points out that even a fake doctor, if given enough case histories, can match a set of symptoms with the most likely disease. So no matter how many green apples our apple AI sees, there will be enough other parameters to make a firm identification. This is where too many dimensions become an issue (page 188).

From page 203 to the end of the book, Domingoes shows how all five AI “tribes” can be combined to yield a more promising version of machine learning to form a foundation for the Master Algorithm. We might not have it yet, but we are getting closer (page 292).

To finish up this long essay, here’s why I think that most AI research should be done under the rubric of machine learning instead of AI. The problem with identifying all of AI with so-called “expert systems” is that AI has fragmented into many different fields, and these vary from those who can show you something working today to those who hope that someday, someway, their brand of AI might come to be. So here are my five “AI Tribes,” ranked from those I consider to be scientists working on firm ground to people who are fantasizing based on flimsy evidence and a lot of wishful thinking:

Machine Learning: What the Master Algorithm is all about. It’s not so much that a computer is AI smart, but that the whole network mining big data will know things about you that even you don’t know (like that you always pick up a case of beer when your spouse sends you for diapers). These are the most grounded and down-to earth of all AI researchers. Their smart robots are here already, out in the Internet, and you can download Weka open-source software or Domingos’s Alchemy machine learning AI environment (page 250)and use it today..

Robot Warriors: Science fiction writer Isaac Asimov’s “robot laws” make it impossible for smart robots to harm humans. Not only were these laws not foolproof (the whole point of his robot stories) but modern robots like military drones exist mainly to kill other humans. The big question today is not if robot soldiers will be able to kill humans – that’s the idea – but if they can do so without a human granting permission. It seems unlikely the delay would be feasible when patrolling mean streets after an invasion. Some fantasize that robot soldiers will only shoot and kill other robots, but it’s hard to see how this “rule” could ever be enforced in a world that still shrugs at wartime rape and torture.

Singularity Soldiers: These guys are in the Ray Kurzweil camp. They call the day when computers become smarter than people the “singularity” (Kurzweil did not invent the term, which computer scientist and science fiction writer Vernor Vinge used in a 1993 essay (page 286), but Kurzweil certainly popularized it). Then perhaps the machines will rise up and crush us, or decide to head for the stars, or keep us as pets. (The best exercise for AI algorithms might be to answer the question of what AI agents will do to humans.)

Consciousness Crusaders: I’m not even sure these guys are separate from the singularity brigade, or if they should be ranked before them. But I put them in their own category because their main goal is not to separate AIs from human society, but sort of to create a separate form of consciousness that we can use to better understand what it is to be human. Researchers here often posit some form of consciousness to other species like apes, dolphins, elephants, or even dogs. This may be so, but I’m not sure who might be interested in creating a robot version of the late, beloved Fido.

Upload Believers: These are the people who think that we can upload the “brains” of dying people into computers and then perhaps download them into other bodies (procured from…?). I remain skeptical that whatever it is that makes us human or conscious (or both) can be separated in any meaningful way from the sensory input system that is our body. It’s possible that you can in some fashion, but I doubt the new entity would be “you.” These guys deal more in religious fervor than scientific rigor, I think.

Population Wars by Greg Graffin

Population Wars: A New Perspective on Competition and Coexistence by Greg Graffin (St. Martin’s Press, 2015), 304 pages.

PW Graffin

Gosh, this is an odd book. I bought it in the bookstore, but a “signed edition” is also available online from several booksellers. The book is billed as a “signed copy” in a bullet sticker on the cover (absent in the bookseller image I used) and a notice on the title page says “This signed first edition…has been specially bound and produced by the publisher.” There is a thick line below this notice and the author’s name is handwritten on the line. Now, I have lots of signed first editions, some of which I obtained at book signings and others that I bought at bookstores after an author’s book signing, but I have to say I have never seen one offered online in quite this way. Either Greg Graffin sits and signs every single book as it rolls off the press, or the signature is “produced by the publisher” in a stamp of some sort. When I buy signed copies in bookstores, I always check to make sure that each signature is a bit different, like humans do. Graffin’s hand is remarkably consistent, but there is a reason that this might be so.

The reason is that Greg Graffin is not only a Cornell University professor with a PhD in Zoology who wrote this book, but the singer and songwriter for the music group Bad Religion. Therefore, I would expect him to be familiar with endless repetitions of his signature, and I know as an author myself that endless repetition often leads to a certain type of uniformity. Now, in the book Graffin says that Bad Religion is a prototypical punk group and all others are pale imitators, but I have to confess I was unfamiliar with his group’s work (my punk experience starts and stops with the Sex Pistols). Judging by the samples available online, I am convinced that Graffin is a much better professor and author than singer and songwriter. But musical tastes vary greatly, so everyone should form their own opinion.

I have to confess that Graffin’s background was the main reason I bought the book. I am a firm believer, along with science fiction author Robert Heinlein, that “specialization is for insects” and people should do as many different things as they can. (That quote is actually a line of dialog in Methuselah’s Children and not a direct Heinlein quote, but there is much evidence of this line of belief in his personal history.)

So what does a singer/songwriter and Cornell professor with degrees in geology and zoology have to add to the perspective on competition and coexistence?

To tell you the truth, the farther I got into this book, the more I had the sneaking suspicion that this book was more the product of academic pressure to publish than the thoughts of a person who just had to express his ideas in print. It makes sense, too: Graffin says he travels all over the world and sings his songs to audiences everywhere. So he has an outlet for expression that few others have, and most would envy. I liked the start and end, but got bogged down a bit in the middle.

I’m being too harsh, I think. After all, I read the whole book without any problem. I’ll just tell you what Graffin says and you can figure out why he wrote it.

Graffin seems to have two main themes. One, as revealed by the subtitle, is that competitions like war and even evolutionary “survival of the fittest” struggles are never completely successful, so coexistence is not only a good idea, it’s inevitable. For example, New York Native Americans like the Iroquois were never wiped out during Colonial expansion, and in fact are more populous and, thanks to casinos, more wealthy than ever before (page 31). (One of my favorite cartoons of all time, which I think I saw in New Yorker magazine, has two Indians and two Pilgrims talking out of the side of their mouths to each other as they get together for Thanksgiving. “Give them alcohol,” says one Pilgrim to the other, “and they’ll give us the whole country.” “Give them gambling,” says the Indian to his partner, “and they’ll give it all back.”)

The second theme, with which I heartily agree, is that “I am more a result of previous circumstances than I am a fulfillment of youthful dreams and willpower” (page 30). I have no doubt that a man whose father was a coal miner can grow up to be vice-president, but I think it’s much more likely if your mom is a banker and you dad is a Harvard law professor. Even so, what we end up doing is more a result of happenstance or coincidence than the force of our will.

Graffin elaborates on page59: it’s very tempting to blame homelessness on a lack of moral fiber or ambition on the part of street people. But the truth is more complex: poverty, sickness, family tragedies, or lack of needed services play a much larger role in your chance of pushing a shopping cart with your belongings than your strength of character. Yet many of the more fortunate lack any sympathy and, perhaps only glancing at their own family history, conclude that “we’re all on our own.”

Graffin keeps coming back to these themes in chapters covering bacteria (people, we now know, are made up of more bacteria than human cells), viruses, and the immune system (that chapter will teach you more about the immune system than you’ve ever learned in school). There is a very long (20+ pages) historical examination of the effects of the French and Indian War and American Revolution on the Iroquois in the Finger Lakes region of New York (this starts on page 160). New York born and raised, I found it all fascinating, especially the part about the Iroquois’ abandonment of the lodge in favor of wood-framed homes, just like the colonials (page 186). But I’m not sure everyone else will enjoy this very detailed exploration of history.

The book picks up again toward the end, when Graffin starts to put it all together. The upper classes always opposed welfare because it removed the “stimulus” of starvation to make the lower classes serve the upper (page 193). The only cure for poverty is to try harder (page 194), right? As Graffin points out on page 201, we accept almost any amount of cheating and foul play to claw our way to the top, where we preen and insist it’s all “natural” and therefore “right” and just. This not only distorts Darwin’s original ideas, but harms society as whole.

I agree with a lot of what Graffin says, but not all. I had a hard time with page 212, where the author examines the story of a woman who drives her car and children into the ocean to kill them and end fourteen years of abuse by her husband and their father. The woman, he claims of page 213, was not acting of her own free will (see also page 220 and 221). I’m willing to grant the downtrodden a helping hand, but I have a hard time with dismissing all aspects of individual responsibility, especially when young innocents are involved.

All of us consider those who we agree with as wise, right up until the point where they diverge from our way of thinking. I think this is true when it comes to our views on education, politics, the homeless, other religions, guns, and many other things. Then we are tempted to try and wipe the others out, when everything in nature tells us we should coexist the best we can.

Of course, that is the entire point of the book….


Machines of Loving Grace by John Markoff

Machines of Loving Grace: The Quest for Common Ground Between Humans and Robots by John Markoff (Harper-Collins, 2015), 376 pages.

MOLG Markoff

The title comes from a poem by 60’s hippy favorite Richard Brautigan. Here, in the “cybernetic ecology” future where no one needs to work, “mammals” are “all watching over by machines of loving grace.” The subtitle is “The Quest for Common Ground Between Humans and Robots” and I guess that’s a good summary.

The author is a Pulitzer-Prize-winning writer about technology and computers for the New York Times. Markoff has been around since 1988, so he’s seen a lot of spotty progress in the fields of robots and Artificial Intelligence (AI) in the past 30 years or so. At least part of his “loving grace” quote is ironic: Markoff is aware of the risks of handing control of crucial human decisions and actions to what are, essentially, machines, although ones that mimic human intelligent (page 8).

Here is the place for my full disclosure: I have done “all but dissertation” in a PhD program for AI at what was the Polytechnic Institute of NY (now part of NYU). So I have my own opinions on AI and robots and minds and consciousness from my (aging) research, and not just what I read in general books on the topic.

Although the subtitle emphasizes “robots” this book considers all types of human-machine interactions. The author talks about driverless cars, sweet-talking smartphone operating systems, and humanoid caregivers designed to provide “companionship” to the elderly. It’s a topic well-worth exploring, and just this week I read more about “Amy Ingram,” an AI entity that you CC on your emails and who will schedule meetings and conference calls while learning your preferences and those of your correspondents. And in Amsterdam, an office building named “The Edge” tells workers where parking spaces are and schedules meeting rooms for them. The chilling aspects of this handover are covered on page 17.

Markoff does readers a real service by dividing AI and robotics researchers into “AI guys” who hope that when the smart robots take over they will perhaps be willing to keep humans are pets and “intelligence augmentation” or “IA guys” (page xii) who don’t mind robots that extend human capabilities, but are never to be allowed to make decisions autonomously. Think of the difference as a giant robot that can roam streets and shoot at targets it determines to be enemies (AI), and a giant exoskeleton suit that protects the soldier within, but never shoots at anything without direct human control (IA).

Lest anyone think that the AI/IA debate is merely a question for the future, consider that by 2018 the Pentagon expects drones and missiles to operate completely autonomously (page 26).

In spite of the willingness to tackle the social issues involved in such things as robot warfare, or abandoning our driving to robot chauffeurs, or assigning care of our parents to robot nurses, much of the book is a history of the development of robotics and AI. Some of it was familiar ground and some of it all new to me. People often forget that Siri was *not* an invention of Apple, but a very ill Steve Jobs decision to purchase the app from SRI. Apple engineers hotly protested they could do better, but Jobs overruled them (page 280). The whole story of Siri (a reversal of a failed “Iris” project at SRI) is told on pages 277 to 305 (!).

Movies and other forms of popular culture are mentioned as well, and these play an important role, I think, in floating test balloons before the general public. However, judging by the box office, it seems to me that many people will accept equality with AI robots (technically androids) as long as they look like hot chicks.

I was a bit surprised that computer and AI pioneer Alan Turing did not get much coverage. Only pages 13 and 14 mention his work, and very briefly. Yet Turing came up with the famous “Turing test” for AI capabilities. Here a human must decide if the conversation they are carrying on (via typewriter or voice of whatever) is with an AI entity or a human being. This game is the source of the title of the movie The Imitation Game, although the move is about WW II code-breaking.

The Turing test, it turns out, is way too easy. I think this might be because Turing, as an ultra-rationale computer scientist, underestimated the ability of “ordinary” humans to take entities they interact with at face value. In other words, if the interface says “Quack!” then there are a lot of people who will shrug and go “I must be talking to a duck.”

I’d like to modestly propose the “Walter test” for AI. In this test, we first clone the AI entity (yes, I know my use of the word “entity” in the AI is somewhat prejudicial, but I do it purposely). Then we separate the two AIs. At one end, we flip a coin to decide if the conversation is to be carried out by the AI entity or a real, living, breathing human being or not. Then we let the other AI entity decide if they are interacting with a human or not. Naturally, this presumes that there is no collusion between the two AIs, so one can’t program the AIs to think “If I use the word ‘possum’ in the third sentence, then I’m your clone AI.” (Yes, I used the word “think” in an AI context on purpose too.) Presumably, the chances are vanishingly small that a real human would play “possum” by chance.

(The whole issue of “artificial consciousness” is handled in this book in what I have come to call the “yeah, so?” stance. In other words, advanced AI robots might actually have as rich an inner consciousness and experience of life as we do, or might just blindly simulate human emotions like caring or anger. But in the end, this argument goes, who cares?)

Markoff handles all of these topics, from military robots like Big Dog to robot caregivers for the elderly, in a very even-handed manner. Regarding the nurses, for example, he points out (page 328) that we as a society have plenty of humans that can care for the elderly, but pay them so poorly and give them so little esteem that it is often a career of last resort or of the extremely altruistic.

The challenge to AI and robotics is outlined on page 257: reach into your pocket, feel around, and pull out a nickel. What you do easily, almost instinctively, is still far beyond the capability of the most advanced robot. True, robotic arms can perform delicate surgery (or toss out party favors at a Silicon Valley billionaire’s party, as on page 242), but always under the control of the human.

On page 336, Markoff sets out crux of the book. The “triple threat” of technology today (robotics, couple with nanotechnology and genetic engineering) might result in the extinction of the human species as we know it. In the end, what we do with AI or IA says more about us than it does about the machines (page 344).

Here’s a final thought. Suppose we fight all our wars using autonomous robots to kill our enemies and help our parents to the bathroom with the same robots re-purposed for the task. Does that say how much we value human life, or how little? Think about it.