THE FUSE IS LIT FOR A POTENTIAL AI EXPLOSION

On the short list of potential existential threats to humanity is an explosive increase in the power of artificial intelligence (AI). 

The people warning us about this argue that the first AI to reach a superhuman level of general intelligence--often referred to as artifical general intelligence or AGI--could rapidly create an even smarter AI, which in turn could create a still smarter AI, etc. This exponential, chain-reaction increase in machine intelligence could quickly leave humanity in the dust, with unpredictable but quite possibly devastating consequences. 

Image credit: TOPBOTS

Skeptics  say that super-smart AGI still lies far in the future, if it will ever be achieved. They also assert that human-positive values could be built into any such system, or that just as we're able to tell current computers and AI systems what to do, we will be able to set goals for and control any artificial general intelligence.

We can reserve a discussion of those issues for a different post.** Today's note is stimulated by the achievement of one of the steps often cited as a lynchpin of a potential artificial intelligence explosion--an AI system designing even smarter "offspring."

James Vincent, writing in The Verge, reports that Google is now using AI to create the "floor plans" for next-generation tensor processing units, or TPUs, chips that Google builds to, among other things, power its most powerful AI systems. 

Reportedly, a new and more potent TPU architecture that would have taken a team of human designers months of intense work to design can now be generated in as many hours by an AI trained to view chip creation as a kind of game. 

Vincent optimistically writes, "The virtuous cycle of AI designing chips for AI looks like it's only just getting started. 

A pessimist might point out that what we've just seen is a speed-up of an important part of AI progress--by an AI--by a factor of more than 700. That's precisely why those pesky AI worry-warts foresee us ever-so-slow humans sooner or later being left wondering what just happened.

REA

---

** You can view some related posts via these links:

zerospinzone: Advanced Artificial Intelligence--Friend or Foe

Article: Google's AlphaZero is now scary smart | OpEdNews

Article: Self-aware artificial intelligence? Coming soon to a robot near you | OpEdNews

zerospinzone: FORGET MOORE'S LAW -- NEVEN'S LAW RULES NOW

HOW SMART IS GOOGLE ASSISTANT?

I've pretty much given up trying to keep track of all the areas where artificial intelligence--AI--has surpassed human capabilities. AI systems now routinely beat even the most expert humans in many once-strictly human skills, from complex games such as chess and Go to more serious capabilities such as reading x-rays. 

Experts keep reassuring us that although AI can beat us at more and more specific tasks or skills, it will be a while, maybe a long while, before an AI system achieves fully human or super-human general intelligence--Artificial General Intelligence or AGI. We'll see.

Hal, what's this piece of music?

Credit: Mixcloud

 

Recently, however, I stumbled on an AI ability that amazes me.

I like classical music and have been listening to it most of my life. Over the years I've made a game of challenging myself to identify whatever piece I happen to hear. I can usually get a piece that I've heard many times, for example one of Beethoven's symphonies or Rachmaninoff's piano concertos. Even if I don't recognize the specific piece, I can usually identify the composer, for example Scarlatti or Stravinsky. And if I can't ID the composer, I can usually glean something about a piece, for example that it's from the Baroque. the Romantic period or the 20th Century.

I'm sure there are many people who can do much better. I remember that my brother's piano teacher claimed he could recognize almost any classical piece from the first few notes. But I doubt that he or any other mere human can do what Google Assistant now does thousands or millions of times per day:

If you hear a classical piece being played, and if you have an Android smartphone, ask your Google Assistant, "What piece of music is this?" Usually within a few seconds, he/she/it/they will tell you the specific piece, the composer, AND THE GROUP, ORCHESTRA OR PERFORMER PLAYING IT.

So not just Vivaldi's Violin Concerto in B-flat, but that it's being played by the Venice Baroque Orchestra, and that the soloist is Giuliano Carmignola.

In order to be able to do this, this system must have "listened to" and "memorized" just about every recording of every piece of classical music (and other musical genres as well), and be able to home in on the specific piece and the specific recording from a few-second-long snippet from anywhere in the piece, all within a few seconds.* As the Google team who developed this system write, ". . . we didn't want people to wait 10+ seconds for a result."

You can read about how Google Assistant pulls off this superhuman feat at this Google Blog post. The gist is that it maps the few seconds of music that it samples through your smartphone  into a series of overlapping 128-dimensional "fingerprints," which are then matched through a two-step process with the most similar fingerprints in the Assistant's constantly updated database of tens of millions of pieces. 

Smart as it is, the system isn't perfect. I ran an experiment asking it to identify 100 classical pieces played on the radio. It correctly identified 97 of them, but it "only" got the group or soloist right on 74 of those. When the Assistant gets something wrong, you can ask it to try again. That brought its piece recognition up to 98 percent, and its ID of the group, orchestra or soloist up to 78 percent. Interestingly, most of its mistakes on orchestras, groups or soloists were on extremely well known pieces, popular workhorses that have been recorded dozens or hundreds of times. 

As is true throughout the AI world, the Google music recognition team is constantly working to make their system even smarter. "We still think there's room for improvement though," they blog, "we don't always match when music is very quiet or in very noisy environments, and we believe we can make the system even faster."

Right now I'm listening to a very familiar piece, The Sorcerer's Apprentice by Paul Dukas. I recognized it right away, and of course so did Google Assistant. But it really pisses me off that the Assistant didn't just identify the piece, but could tell that it was the version recorded by the Montreal Symphony Orchestra led by Kent Nagano. Damned showoff.

-----

*To be fair, it appears that the Shazam app, now owned by Apple, has pretty much the same capabilities, although it seems a tad slower to me.

GOOGLE CLAIMS QUANTUM SUPREMACY--THIS IS BIG!

Just six weeks ago I posted about Google quantum AI guru Hartmut Neven predicting that quantum computing would grow at an unheard-of doubly exponential rate, and that the long-sought benchmark of quantum supremacy would be reached sometime this year.

Quantum computer core

Credit: Flickr

True to Neven's predictions, word has filtered out that Google has submitted a scientific paper reporting the first demonstration of a quantum computer solving a problem that even the most powerful classical computer can't manage--the hallmark of quantum supremacy. Reportedly, the problem Google's 54-qubit computer solved in 200 seconds would have taken a supercomputer 10,000 years to do.

One caveat--the problem was a very specific task known to be particularly well suited to a quantum computer. Much more work is needed before quantum computers will be able to tackle a full range of real-world problems. However, those developments will almost certainly happen much sooner than most people imagine.

This milestone is important in itself, meaning that scientists in every field, cryptographers, AI researchers, etc., will soon be able to tackle tasks that were previously impossible. However, what really demands everyone's attention is that easily-missed prediction, now known as Neven's law, that progress in quantum computing is going to unfold at a doubly exponential rate.

"To our knowledge," the Google team writes, "this experiment marks the fi rst computation that

can only be performed on a quantum processor. Quantum processors have thus reached the regime of quantum supremacy. We expect their computational power will continue to grow at a double exponential rate."

You can get a sense of what this astonishing rate of change means in my earlier post, "Forget Moore's Law. Neven's Law Rules Now." The bottom line is that we can expect as much progress in quantum computing in the next five or six years as we've seen in the digital world over the past five or six decades. After that point, all bets are off. If you think that the digital revolution has been earth shaking, just wait for the quantum revolution.

As Nevens says, "It looks like nothing is happening, nothing is happening, and then whoops, suddenly you're in a different world."

Well, we're in that new world now. It's going to make words to describe the rate of change, such as "breathtaking," "jet propelled" or "explosive" seem far too slow.

-----

Robert Adler

GOOGLE'S ALPHAZERO IS NOW SCARY SMART

What would you think if you taught your child the rules of chess at breakfast and found that by lunchtime she had beaten the world champion? Awe? Parental pride? A bit of fear perhaps?

That's essentially what just happened at Google's DeepMind subsidiary in London. They created an ultra-powerful game-playing computer system called AlphaZero based on a neural network capable of deep learning through reinforcement.

Unlike other chess-playing programs--which have outperformed humans since IBM's Deep Blue beat the human world champion, Gary Kasparov, in 1997--AlphaZero was not pre-programmed with any specialized knowledge or expertise about chess. It was simply given the rules of the game and allowed to learn by playing against itself.

Four hours later AlphaZero crushed the World Computer Champion, Stockfish, with 28 wins and zero losses in a 100-game tournament (the remaining games were ties).

 World Champion Gary Kasparov struggles against IBM's DeepBlue, 1997

Credit: iChess.net

British chess expert Colin McGurty sums up AlphaZero's achievement:

The AlphaZero algorithm developed by Google and DeepMind took just four hours of playing against itself to synthesise the chess knowledge of one and a half millennium and reach a level where it not only surpassed humans but crushed the reigning World Computer Champion Stockfish 28 wins to 0 in a 100-game match. All the brilliant stratagems and refinements that human programmers used to build chess engines have been outdone, and like Go players we can only marvel at a wholly new approach to the game.

Other chess experts describe AlphaZero's play as "divine," or "from another galaxy." 

As if one superhuman feat were not enough, the AlphaZero team used the same artificial intelligence (AI) system to tackle the games of Go and the Japanese chess game, Shogi. It took AlphaZero just two hours of play against itself to surge past Elmo, the Shogi Computer World Champion, and all of 8 hours to surpass AlphaGo (another DeepMind program), which itself dethroned the human Go champion, Ke Jie, earlier this year.

So, to summarize, in less than a day, starting as a blank slate knowing nothing more than the rules of the games, and simply by playing against itself, AlphaZero reached a superhuman level of play in three abstract games that have challenged humans for millennia. Not a bad day's work.

And just in case you're thinking that AlphaZero reached these superhuman levels simply by calculating faster than any other computer, that's far from the case.  It is blindingly fast compared to humans--for example searching 80 thousand chess positions per second, but it is tortise-slow compared to other chess-playing systems. Stockfish, which AlphaZero completely dominated, searches 70 million positions every second. The system's creators explain, "AlphaZero compensates for the lower number of evaluations by using its deep neural network to focus much more selectively on the most promising variations -- arguably a more 'human-like approach to search."

In recent years some very smart people including Bill Gates, Elon Musk and Stephen Hawking have warned about the threat posed by out-of-control artificial intelligence. While the superhuman learning and game-playing of AlphaZero seem benign, and ubiquitous, mostly invisible AI applications help us every day in a huge variety of areas, there are red flags raised by robo-cops and soldiers, vital infrastructure managed by AI, increasingly capable and autonomous robots that may replace most workers, and the potential for super-intelligent AI creations that may not have the interests of humans at heart. AlphaZero, for example, could equally well learn to "play" at a superhuman level at politics, finance or war.

Gates and others emphasize that we need to figure this out before such intelligences emerge because once they do, like AlphaZero, they could leave us in the dust within a few hours.

So if your child became the world champion chess player after four hours of play, wouldn't you be scared? I would.

-----
You can read the scientific paper describing AlphaZero's accomplishments here.
-----
If you enjoyed this post, please sign up to follow or receive email notices from zerospinzone.blogspot.com (see upper right column).

---

Ahora AlphaZero de Google está peligrosamente inteligente

¿Que pensaría si le hubiera enseñado a su niña las reglas de ajedrez a las ocho de la mañana y descubriera que a mediodía ella ha ganado contra el campeón del mundo? ¿Asombro? ¿Orgullo parental? ¿Quizás un poco de miedo?

Eso es esencialmente lo que pasó en DeepMind, un subsidiario de Google en Londres. Ellos crearon un sistema informático ultra poderoso para jugar juegos, se llama AlphaZero, lo que implementa una red neuronal artificial capáz de realizar un aprendizaje profundo a través del refuerzo autónomo.

Es importante darse quenta que programas que juegan al ajedrez han superado a los humanos desde entonces Deep Blue, de IBM, lo ganó el campeon del mundo humano, Gary Kasparov, en 1997. Deep Blue y todas los programas aun mas poderosos que han sido desarrolados después, empiezan con una gran cantidad de conocimiento y pericia dado por expertos humanos. Por el contrario, AlphaZero no fue programado con algun conocimiento especializado, ni pericia. Nada mas supo las reglas del juego, y recibió instrucciones de aprender a través de jugar contra sí mismo.

Cuatro horas más tarde, AlphaZero le aplastó el Campeón Mundial de Ajedrez por Computadora, Stockfish, con 28 victorias y zero derrotas en un torneo de 100 juegos (los juegos restantes fueron atados).

El experto de ajedrez británico, Colin McGurty, resume lo que logró AlphaZero asi:

El algoritmo AlphaZero desarrollado por Google y DeepMind necesitaba nada más cuatro horas de juego contra si mismo para sintetizar el conociemiento de ajedrez de un milenio y medio y llegar a un nivel donde el no solo superó los humanos pero aplastó el reinante Campeón Mundial de Computadora, Stockfish, 28 victorias a 0 en un torneo de 100 juegos. Todas las estratagemas y refinamientos que los programadores humanos usaban para construir engines de ajedrez han sido superados, y como los jugadores de Go, no podemos hacer más que maravillarnos con un enfoque totalmente nuevo para el juego.

Otros expertos de ajedrez describen el juego de AlphaZero como “divino,” o “de otra galaxia.”

Como si una hazaña sobrehumana no fuera suficiente, el equipo de AlphaZero usaron el mismo sistema de inteligencia artificial (AI) para tratar de dominar el juego de Go y el juego de ajedrez japonés, Shogi. AlphaZero necesitaba nada más dos horas de juego contra si mismo para abrumar a Elmo, el Campeon Mudial de Computadora de Shogi, y un totál de 8 horas para superar al AlphaGo (un otro programa de DeepMind), lo cual destronó el campeon humano de Go, Ke Jie, a principions de este año.

Entonces, para resumir, en menos de un día, empezando como una pizarra en blanco, sabiendo solo las reglas de los juegos, y simplemente jugando contra sí mismo, AlphaZero alcanzó un nivel de juego sobrehumano en tres juegos abstractos que han desafiado los humanos por milenia. No está mal para el trabajo de un día.

Y si estás pensando que AlphaZero alcanzó estos niveles sobrehumanos simplemente por calculando más rapidamente que alguna otra computadora, no es la verdad. AlphaZero si es deslumbrantemente rapido comparado con los humanos—por ejemplo buscando 80 mil posiciones de ajedrez cada sugundo, pero es lento como una tortuga comparado con otros sistemas que juegan al ajedrez. Stockfish, lo cual AlphaZero dominó por completo, busca 70 millones de posiciones cada segunda. Los creadores del sistema explican, “AlphaZero compensa el menor número de evaluaciones mediante el uso de su red neuronal profunda para enfocar mucho más selectivamente en las variaciones más prometedoras – discutiblemente un enfoque mas humano a la busqueda.

En los años recentes, algunas personas muy inteligentes, incluso Bill Gates, Elon Musk y Stephen Hawking han advirtido sobre la amenaza de inteligencia artificial fuera de control. Es verdad que el aprendizaje y el juego de AlphaZero parecen benignos, y hay un monton de aplicaciones de AI, ubicuos y en su mayoría invisibles, que nos ayudan diaramente y en una variedad grande de maneras. Pero, al mismo momento, debemos de notar la posibilidad de soldados y policías robóticos, infraestructura vital administrada por AI, robots cada vez más capaces y autónomos que pueden reemplazar la mayoría de los trabajadores, y el potencial para creaciones AI de inteligencia sobrehumano que no tendrían los intereses humanos “en sus corazones.”

AlphaZero, por ejemplo, podría “jugar” a un nivel sobrehumano a la política, las finanzas, o la guerra.

Gates y otros enfatizan que necesitamos resolver este asunto antes de que una inteligencia de este tipo surja, porque, una vez que aparezca, nos puede dominar o superceder entre pocas horas.

Entonces, si su hija se convirtió en campeona mundial de ajedrez después de cuatro horas de juego, ¿no estaría asustado? Yo si estaría.