Magic iPod

Everyone has been passing around the Magic iPod this month. First Deep Blue beat Kasparov, then AlphaGo beat Lee Se-dol, and now we have Magic iPod taking down Girl Talk.

When you read stories about how artists come up with mashups (finding works with compatible BPM and keys, among other things), or how the Swedish pop factory mad scientists like Max Martin conjure pop hits, it seems inevitable that in our lifetime we'll have algorithms creating real pop hits.

How such work is received by a human audience is about more than its intrinsic qualities, however. In an objective competition like a game of Go, or when considering a mashup which is simply the synthesis of existing creative works, I suspect humans will be comfortable with acknowledging the achievements of an algorithm.

With original creative works, however, like music, novels, movies, I suspect humans will recoil from even intrinsically appealing creation if it was written by a computer program. Call it some variant of the uncanny valley effect.

We have a romantic attachment to human creation, and it may take a generation of people passing on before we overcome that cultural aversion. When a waiter places a beautiful dish in front of you at a restaurant, we like to imagine that a chef toiled over the plate in the kitchen, conjuring that beautiful, delicious entree from raw ingredients, fire, and ingenuity. When we read an engrossing novel, we picture a tortured writer banging on an old typewriter in a cabin by the sea, stopping from time to time to put out a cigarette and gaze out the window at the ocean waves trying to claw up the gentle slope of the beach.

When Beyonce drops Lemonade or any one of her jaw dropping awards show performances on an unprepared world, I like to believe the work was birthed from what is surely a vagina with mystic powers, belonging as it does to our modern icon of feminism and black empowerment.

It's not quite as appealing if the truth was that an algorithm finished processing in some computer lab somewhere. A progress bar on a monitor finally reaches 100%, and a file is deposited into a directory.

That's why if humans ever comes up with algorithms that are capable of creating popular works of culture, it's financially wise for the creators to claim the credit themselves, at least until many years of critical and popular embrace have accumulated. Then, and only then, spring the truth on the world.

We live in a Skinner box, and it was of our own making.

RankBrain

For the past few months, a “very large fraction” of the millions of queries a second that people type into the company’s search engine have been interpreted by an artificial intelligence system, nicknamed RankBrain, said Greg Corrado, a senior research scientist with the company, outlining for the first time the emerging role of AI in search.
 
RankBrain uses artificial intelligence to embed vast amounts of written language into mathematical entities -- called vectors -- that the computer can understand. If RankBrain sees a word or phrase it isn’t familiar with, the machine can make a guess as to what words or phrases might have a similar meaning and filter the result accordingly, making it more effective at handling never-before-seen search queries.
 
[...]
 
RankBrain is one of the “hundreds” of signals that go into an algorithm that determines what results appear on a Google search page and where they are ranked, Corrado said. In the few months it has been deployed, RankBrain has become the third-most important signal contributing to the result of a search query, he said.
 
[...]
 
So far, RankBrain is living up to its AI hype. Google search engineers, who spend their days crafting the algorithms that underpin the search software, were asked to eyeball some pages and guess which they thought Google’s search engine technology would rank on top. While the humans guessed correctly 70 percent of the time, RankBrain had an 80 percent success rate.
 

More on RankBrain here.

Machine learning is advancing fast. At its best it feels a bit like magic, and it's endlessly malleable. Think it's missing something of importance? Add it as a factor, or tune it up.

I suspect in my lifetime we'll have machine learning so good it will be largely incomprehensible to me. That is, it won't be understandable by using analogies to how humans think because it will be its own form of intelligence.

A cookbook from IBM's Watson

Robots taking all the jobs, cooking edition:

Steve Abrams, the director of IBM’s Watson Life research program, told Quartz that Watson scanned publicly available data sources to build up a vast library of information on recipes, the chemical compounds in food, and common pairings. (For any budding gastronomers out there, Abrams said Wikia was a surprisingly useful source.) Knowledge that might’ve taken a lifetime for a Michelin-starred chef to attain can now be accessed instantly from your tablet.
 
What separates Watson from the average computer (or chef) is its ability to find patterns in vast amounts of data. It’s essentially able figure out, through sheer repetition, what combinations of compounds and cuisines work together. This leads to unusual pairings, like Waton’s apple kebab dish, which has some odd ingredients: “Strawberries and mushrooms share a lot of flavor compounds,” Abrams said. “It turns out they go quite well together.”
 

The researchers are publishing a cookbook with recipe ideas from Watson, and it releases this Tuesday: Cognitive Cooking with Chef Watson: Recipes for Innovation from IBM & the Institute of Culinary Education. I have not read the book, but some of the recipes sound intriguing (“Belgian bacon pudding, a desert containing dried porcini mushrooms”) while others sound, at best, like clever wordplay (“the shrimp cocktail, which is a beverage with actual shrimp in it”). Regardless, I'm purchasing a copy just out of sheer curiosity. Let's hope they turn this resource into an app or service instead of a book, I blame Watson's vanity for wanting this in the outdated format of a book.

To the extent that standout recipes and flavor pairings are a matter of pattern recognition, there's no reason a computer, with its infinitely more scalable hardware and software for that purpose, couldn't match or exceed a human. And, so, a variant of the infinite monkey theorem: given enough time, a computer will write the French Laundry cookbook (and win a third Michelin star).

To be clear, I'm okay with this. I just want to eat tasty food, I'm fine with employing computers to come up with more amazing things to feed me.

For now, however, the computer still requires a human to actually prepare the recipe. In a true demonstration of how far artificial intelligence has progressed, no sufficiently advanced computer wants the drudgery of life as a line chef. Better profits in cookbooks than restaurants anyway.

A new cooking show concept already comes to mind: Top Freestyle Chef. Like freestyle chess, in freestyle cooking competitors would consist of a human or a human consulting with a computer. I am ready to program this into my DVR already, as long as they don't replace Padma Lakshmi with a robot host. I'm as big a fan of artificial intelligence and robots as the next guy, but I think we're a long way from replacing this.

Opaque intelligence

Alex Tabarrok writes about what he calls opaque intelligence.

It isn’t easy suppressing my judgment in favor of someone else’s judgment even if the other person has better judgment (ask my wife) but once it was explained to me I at least understood why my boss’s judgment made sense. More and more, however, we are being asked to suppress our judgment in favor of that of an artificial intelligence, a theme in Tyler’s Average is Over. As Tyler notes notes:

…there will be Luddites of a sort. “Here are all these new devices telling me what to do—but screw them; I’m a human being! I’m still going to buy bread every week and throw two-thirds of it out all the time.” It will be alienating in some ways. We won’t feel that comfortable with it. We’ll get a lot of better results, but it won’t feel like utopia.

I put this slightly differently, the problem isn’t artificial intelligence but opaque intelligence. Algorithms have now become so sophisticated that we human’s can’t really understand why they are telling us what they are telling us. The WSJ writes about driver’s using UPS’s super algorithm, Orion, to plan their delivery route:

Driver reaction to Orion is mixed. The experience can be frustrating for some who might not want to give up a degree of autonomy, or who might not follow Orion’s logic. For example, some drivers don’t understand why it makes sense to deliver a package in one neighborhood in the morning, and come back to the same area later in the day for another delivery. But Orion often can see a payoff, measured in small amounts of time and money that the average person might not see.

One driver, who declined to speak for attribution, said he has been on Orion since mid-2014 and dislikes it, because it strikes him as illogical.

One of the iconic moments from Hitchhiker's Guide to the Galaxy is when a supercomputer finally finishes computing, after 7.5 million years, the answer to the ultimate question of life, the universe, and everything, and spits out 42. Perhaps that is how far beyond our understanding a super-intelligent AI will be. We may no more understand them than a snail understands humans. Defined that way, opaque intelligence is just artificial intelligence so advanced we don't understand it.

Someday a self-driving car will make a strange decision that will kill someone, and the software will be put on trial, and despite all the black box data recovered we may have no idea what malfunctioned. Sometimes my iPhone randomly crashes and reboots, I couldn't begin to tell you why.

I'm waiting for the dystopic sci-fi movie that postulates an armageddon scenario much more likely than Skynet in Terminator. That is, rather than waste time building cyborg robots to hunt us down, a truly super-intelligent AI that wanted to kill off humans could just simultaneously order a million self-driving cars to speed headlong into each other, all the planes in the world to plunge into the ground, all our nuclear reactors to melt down, and a dozen other scenarios far more efficient than trying to build humanoids that walk on two legs.

Not as visually exciting enjoyable as casting Arnold, though. In a way, it's reassuring that for all the supposed intelligence of Skynet, it sends back a Terminator that still has a terrible Austrian-accented English, as if artificial speech technology was the one technology that failed to keep up despite AI making leaps as complex as gaining consciousness.