The Age of Spiritual Machines
Is consciousness an algorithm? Computers are not as smart as we think. Yesterday I tried to print photos from a memory. The machine at the pharmacy told me that my memory was blank. I knew I had at least 2,000 photos on the memory, but c’est la vie. I asked the young person for help and she treated me like a senile delinquent. Obviously I had no idea how to use the photo machine. She showed me how to insert my memory in the machine and push a button. The message told me that my memory was blank. An electronic caprice denied me access to my photos of elephants in Thailand. So it goes.
A bit later on the same day, I had to reserve a medical appointment on the internet. The system was down. I called the 800 number. The secretary on the telephone put me on hold. 20 minutes later she told me the system was down.
A new ransomware attack is infecting airlines, banks, and utilities across Europe. The cyber attack that apparently began in Ukraine has specialists puzzled. The most severe damage is being reported by Ukrainian businesses, with systems compromised at Ukraine’s central bank, state telecom business, municipal metro and and Kiev’s Boryspil Airport. Systems were also compromised at Ukraine’s Ukrenego electricity supplier, although at present the power supply was unaffected by the attack.
That things sometimes don’t work and go bump in the night is nothing new. I don’t expect the technology to work all the time. Different cultures respond differently to such failures.
In 1984, I was buying some pencils at an art supply store in San Jose, California when the lights went out in the store. It was a crisis. The girls running the store called everyone to the front of the store and asked everyone to leave quietly. Without electricity there was no way to do business. A week later I was in a market place in India when the lights went out. No one blinked. Prices were tallied and sales were registered with pencil and paper. But that was long ago.
Today we place so much importance on electronic communication that it has become obligatory. Many more aspects of our lives depend on the caprices of electronics. From shopping, electronic checkins, online hotel reservations, buying things on Amazon.com, even keeping up with friends on social media, we are slaves of the machine, prisoners of the internet.
In 1984 I interviewed Hubert Dreyfuss at Berkeley on the subject of Artificial Intelligence. He felt that no computer would achieve the level of artificial intelligence needed to defeat a human chess champion in the 20th century. He was almost right. In 1997 an IBM supercomputer called Deep Blue defeated chess champion Garry Kasparov in a series of schess matches. Dreyfuss argued that computer intelligence is “rule-based,” and can reach competence, but not expertise. True human expertise, he felt is intuitive and situational, based on thousands of experiences. No computer could ever match human expertise, he felt.
The idea in those days, mostly advanced by Marvin Minsky and computer scientists at MIT, was to create what he called “neural networks.” Enough microprocessors linked together might be able to imitate the thinking capacity of the human brain.
Writing in the 1990s in “Are We Spiritual Machines,” eccentric inventor and futurist Ray Kurzweil argued that computers will reach and surpass human intelligence. The engineering notion called “Moore’s Law” has correctly predicted that both machine memory and speed in IT technology doubles ever 2 years. Given exponential technological change, advances in “Artificial Intelligence” will be much more dramatic than previously felt.
For Kurzweil, change does not occur at a fixed rate, but is exponential, contrary to the common-sense “intuitive linear” view. So we won’t experience 100 years of change in the 21st century at a fixed rate: Change will occur exponentially; systems will be affected dramatically in what will be more like thousands of years of transformation in a short amount of time. We will witness exponential change in population growth, climate change, and especially Information Technology:
“Within a few decades, machine intelligence will surpass human intelligence, allowing nonbiological intelligence to combine the subtleties of human intelligence with the speed and knowledge sharing ability of machines.”
This is certainly the stuff of science fiction: at least it was back in the 1990s. But Kurzweil was right about exponential growth, especially in an unforseen area: the internet and “cloud” computing. If Minsky’s idea of somehow linking a bunch of computers together to create a “neural network” was naive and immaturely conceived, the latest work in machine intelligence may be the realization of his dream. It is no longer necessary to physically link experimental super-computers in a fanciful network for the exclusive use of scientists: we have GOOGLE.
In a new book, Machine, Platform Crowd, Machine Over Mind In A New Economy, a new generation at MIT give us a glimpse of the future of technology. MIT’s Andrew McAfee and Erik Brynjolfsson examine an exponential digital-powered shift. We find ourselves at a crossroads, where we will be forced to rethink the integration of minds and machines, of products and platforms, and of the core and the crowd. The new subtlety in artificial intelligence has massive implications for how we run our companies and live our lives.
Crowd-sourced machine intelligence coupled with the exponential growth in memory and design drive a new generation in artificial intelligence.
Consider Google Translate . Traditionally, language is one of the most complex human phenomenon. A proper translation may be incredibly subtle; that’s why jokes, puns, poems, and even diplomacy is often “lost in translation.”
Google Translate is not merely a dictionary translation. It uses subtle algorithms to reach its goals. While primitive in its inception, over the years, Google Translate has gotten significantly better at giving its users (relatively) legible translations for most commonly used languages. Still far from perfect, in 2014 Google announced a new initiative that aims to get more input from its users to improve its translations.
The Google Translate Community, open for everyone, has been giving users the option to offer their own translations and validate current translations. The millions of Google users constantly offer human input into the subtlety of the translations made through algorithms. As the human translation experts influence the patterns, the deep neural nets become more capable of offering correct translation.
The innovation is in the way computer models are influenced by human input multiplied millions of times through crowd-sourcing and wikis. Language translation is far from perfect, but it is now serviceable enough to replace human translation services.
A new Google application called Crowdsource has quietly appeared on Google Play, asking users to perform brief tasks that will help improve the quality of Google services like Maps, translation, image transcription, and more. As millions participate, machine intelligence learns to perfect tasks once only available to humans.
Perhaps game-playing is more revealing of the exponential advances in artificial intelligence. Human chess champions were defeated by machine intelligence 20 years ago; the latest human thinkers to be crushed by computers are players of the board game “Go.”
“Go” is interesting because it is a far more subtle game than chess. There are millions and millions of potential moves impossible to analyze through mere brute force and data bases. But with the help of thousands of Go masters giving input on which positions are more “favorable” than others, algorithms are developed which couple grandmaster intuition with brute number-crunching.
Here’s Scientific American on the differences between the DEEP BLUE machine that busted Gary Kasparov and the new AlphaGo system of AI that defeated the world’s greatest Go-master Lee Se-dol:
Deep Blue represented a triumph of machine brawn over a single human brain. Its success was almost completely predicated on very fast processors, built for this purpose. Although its victory over Kasparov was a historic event, the triumph did not lead to any practical application or to any spin-off. Indeed, IBM retired the machine soon thereafter.
The same situation is not likely to occur for AlphaGo. The program runs on off-the-shelf processors. Giving it access to more computational power (by distributing it over a network of 1,200 CPUs and GPUs) only improved its performance marginally. The feature that makes the difference is AlphaGo’s ability to split itself into two, playing against itself and continuously improving its overall performance. At this point it is not clear whether there is any limitation to the improvement AlphaGo is capable of. (If only the same could be said of our old-fashioned brains.) It may be that this constitutes the beating heart of any intelligent system, the Holy Grail that researchers are pursuing—general artificial intelligence, rivaling human intelligence in its power and flexibility.
After losing the second match to Deep Mind, Lee Se-dol said he was "speechless" adding that the AlphaGo machine played a "nearly perfect game".
The two experts who provided commentary for the YouTube stream of for the third game said that it had been a complicated match to follow.
They said that Lee Se-dol had brought his "top game" but that AlphaGo had won "in great style".
The BBC reported: “The AlphaGo system was developed by British computer company DeepMind which was bought by Google in 2014.
It has built up its expertise by studying older games and teasing out patterns of play. And, according to DeepMind chief executive Demis Hassabis, it has also spent a lot of time just playing the game.
"It played itself, different versions of itself, millions and millions of times and each time got incrementally slightly better - it learns from its mistakes," he told the BBC before the matches started.” (http://www.bbc.com/news/technology-35785875)
Game-playing computers, of course, are a fun curiosity. But what are the implications of more and more advanced “machine intelligence?”
In hospital emergency rooms, health workers are encouraged to follow “checklists,” that enable patients to be channelled through the system properly according to their symptoms. In a flu epidemic it is vital to sort potential carriers of the virus from ordinary cold sufferers. Chest pain checklists help doctors understand whether a patient suffers from heart burn or a heart attack. Some of this work can be automated. As more and more symptoms are handed over to computers, doctors have more time to spend on healing.
But there’s always a certain amount of “mission creep.” Here in Mexico, appointments were always made personally with the local doctor. The health system was recently computerized, making it possible to make appointments online. What worked fine when there were a limited number of people in the system, however, hasn’t worked well after it was made mandatory. There’s a tendency to try out a system to see if it works and when it does to offer it online. Once it’s really up and running, it’s made obligatory.
Online checkin for airlines was once a luxury; now its mandatory. As humans and machine work together to get a faster throughput in a more efficient society, more and more tasks are turned over to machines. A simple problem of developing a photo becomes entangled in technology when the chip-reader at the pharmacy finds my memory is blank. Things don’t always work as well as they are supposed to. That doesn’t seem much, but as we hand the work of intelligent humans over to intelligent machines, it gets stickier.
People everywhere were once flexible in their responses to everyday situations. Like the folks in India who went on selling cloth in their shops even when the lights went out, we were able to work things out with pencils on paper. But a relatively large number of young people today seem to have lost the flexibility to deal with quotidian situations. Wherever we go, whatever we do, shopping, going to the movies, at the hotel, in the restaurant, at the doctor’s office there’s a computer controlling the transaction. We are all accustomed to this; life seems easier because of all the technology.
But what happens when the computers that control our transactions are themselves controlled by accounting software? What happens when the policies driving transactions are controlled by algorithms? The other day in Mexico City I was dropped from the hospital data bank. The doctor explained that since my heart condition was no longer an emergency, the computer algorithm determined that I should no longer be in the system and was red-flagged and removed. Happy as I was to learn that I was no longer an emergency, I was surprised to learn that the algorithms driving policy ran the computer.
Many policy-making decisions are now determined by computer models which determine how to make things more efficient. As the technology itself configures the technology, we find machines designing machines, clouds and crowds determining the algorithms that drive the system, reducing the role of humans.
Young people are trained to perform computation without comprehension. They merely operate the system at the end-user point of sale. It seems that as machines become more artificially “intelligent” the culture itself becomes less mindful.
MIT School of Management researchers McAfee and Brynjolfsson point out in their new book, Machine, Platform, Crowd, that the world of artificial intelligence is transforming the economy. The disruption is plain to see. Our ways of shopping and doing business are changing fast in the online economy. Where shopping malls were once the pride of America, in the last decade, more than 20 percent have closed. Where people once met in the marketplace, they are finding it cheaper to stay home and shop on line, feeding giants like Amazon.com.
I often feel frustrated that given all the telephones and communication platforms, it seems impossible to actually have a conversation with someone. We have Skype, WhatsApp, FaceBook, Instant Messenger, E-mail, cellphones and landlines. But every time I try to call someone I get a busy signal or a voice-mail message. People feel less connected. Perhaps this is because of the fact that almost two-thirds of millennials don’t have landline phones.
The tourist and transportation business is also under transformation. Thanks to the new “gig economy” innovations like Airbnb and Uber, people are avoiding hotels and taxis.
Instead of malls, traditional stores, hotels, and taxis, now we have “platforms”—organizations without inventory and sometimes without much of an organization.
These platforms are technologically driven “brands” and are gradually becoming more competitive than brick-and-mortar companies as technology and artificial intelligence becomes more dominant.
Artificial Intelligence sounds great and futuristic as long as we’re talking about chess-playing machines. But what happens when machines replace experts? What happens when humans are unable to question systems created by computer models controlled by algorithms?
It sounds like a dystopian nightmare. But are we really so far away from the kind of future described by Isaac Asimov? Among the facets of our brave new world are the algorithmically driven “automatic decisions,” by which Amazon cross-recommends products to shoppers. You’ve probably seen ads driven by these algorithms on your facebook page. I find myself bombarded with ads that push cheap airfare prices that correspond to my Google searches; instead of bright young madison avenue admen, algorithms study my web activity and offer me proposals and projects based on my psychological profile.
Besides virtual reality, the self-learning algorithms of artificial intelligence have taken over manufacturing and agricultural sectors as well. Manufacturing jobs that haven’t gone to China are done by robots. In this video, fully automated car factories produce BMWs with little human input.
https://www.youtube.com/watch?v=VpwkT2zV9H0 But car manufacturing is not the only area where robots are used. History teachers in California make sure that young people know the story of Cesar Chavez who used Gandhian nonviolent resistance to organize the farm-workers who harvested lettuce in the 1970s. Now, robots harvest lettuce.
https://www.youtube.com/watch?v=_i62juq8Euk. There’s no need for the lettuce farm-workers anymore. Cotton-picking is also mechanized as is much of agriculture.
We may assert that unlike machines, we have free will. But our freedom has its limits. McAfee and Brynjolfsson point out that our new information economy will be ruled by the elite one percent who own and control everything. In this world, machine learning, AI, and robotics, will have far more disruptive effects, displacing human labor wherever possible, while the winning firms of the near-term future will leverage these shifts to “bring together minds and machines, products and platforms, and the core and the crowd very differently than most do today.”
According to McAfee and Brynjolfsson, while people may be allowed some input, algorithmically driven “automatic decisions,” will drive the future. The leading companies of the second machine age may look very different from those of the industrial era,” write the authors, “but they will almost all be easily recognizable as companies.”
If machines can think, where does that leave consciousness? Are machines really thoughtful? Or as Ray Kurzweil put it, “Are we spiritual machines?”