|
|
Musings Report #32 8-6-16 Will Super-Intelligent AI Take Over the World?
You are receiving this email because you are one of the 500+ subscribers/major contributors to www.oftwominds.com.
For those who are new to the Musings reports: they are basically a glimpse into my notebook, the unfiltered swamp where I organize future themes, sort through the dozens of stories and links submitted by readers, refine my own research and start connecting dots which appear later in the blog or in my books. As always, I hope the Musings spark new appraisals and insights. Thank you for supporting the site and for inviting me into your circle of correspondents.
Will Super-Intelligent AI Take Over the World?
The narrative that artificial intelligence (AI) will soon exceed human intelligence and then expand far beyond our meager smarts has been around for many years, popularized by a media-savvy handful of techno-seers.
This article does an excellent job of summarizing the narrative and introducing a few of the skeptics' points of view.
AI Revolution 101: Our last invention, greatest nightmare, or pathway to utopia?
I sent the article to longtime correspondent Don S. and asked for his feedback. Don and I have been exchanging emails on Nature's immense potential for rapid evolutionary adaptation, and what this might mean for a human status quo that is clearly on an unsustainable pathway.
Regarding the techno-enthusiasts' confidence that AI will soon exceed human capabilities, Don asked: what if humans don't want whatever it is that AI has to offer us?
Here are his comments:
"Here is a radical thought for you. Suppose that the future requires us to steadily prune away what is not essential. I was struck today how essential selection has been in producing the humans we are today.
The physical repertoire we have is ideally suited to a certain kind of life on earth. Many of our problems arise because we are no longer living the life we are designed to live. For example, if you listen to doctors, they will tell you to go to bed at sundown and get up at sunrise. Working on computers at 9:45 pm, as I am doing now, upsets our circadian rhythms. Likewise, our bodies need to experience changes in temperatures, but we set thermostats and then we miss the signals we are supposed to get.
Toby Hemenway wrote an essay after visiting the Battle of the Little Big Horn site.
He says in here somewhere that there couldn't be peace between the United States and the plains Indians because 'there was nothing the US had to give them which they wanted'. My question to you: How could it be that a mighty industrial power, which had recently mobilized enormously destructive forces to win the Civil War, had nothing that the Indians wanted?
Here is the radical thought: Suppose that our destiny is to evolve into the kind of people who are not enticed by anything the Artificial Intelligence crowd have to offer? Its a little like the Amish...they don't want what we have.
Similarly, several years ago I was waiting to pick my wife up at a craft center. I had talked to the young woman who worked the front desk before, and knew that she was a potter. I asked her if she was going to the big pottery affair over the weekend. I was shocked when she said 'No, I need to go to church'.
In my culture, going to church could always be put off to another day, while a special event wouldn't come around again. I was struck by the realization that the young woman was finding something in church that could not be found at a pottery workshop.
Perhaps we are headed for a time when people will be unable to watch spectator sports on television because 'I have to take my children and cane poles down to the creek to fish,'"
Don's question bypasses the technical questions about computational limits and goes straight to heart of the role of computational intelligence in the human economy, society and culture.
The techno-enthusiasts assume AI will be uncontrollably powerful. They assume the rapid and endless expansion of AI's "intelligence" (however we define the spectrum of traits we crowd into the word "intelligence") is a given, and that humanity could either benefit from this or, if things go amiss, be destroyed by an unplanned expansion of nanobots or other technology.
Don asks: what if we don't want what AI has to offer? There are many reasons why we may be choosy about which kinds of AI we want to unleash.
In other words, the techno-enthusiasts assume that AI is just another item in the endless list of new technologies that we'll embrace because it's so compellingly advantageous.
But advantageous to whom, and to what ends?
The potential ramifications are not just some sci-fi scenario or economic disruption; there are political consequences as well. Suppose the super-AI is tasked by the state with "eliminating terrorism by any means available"? Who gets to program the definition of terrorism the machines will follow? Who (if anyone) gets to file an appeal before the super-AI launches a drone strike or audit the super-AI network's choices and performance?
My interest in AI stretches back over 30 years, and I've read many books on the topic and tracked the many false hopes that have fueled techno-fantasies.
I suspect that the AI enthusiasts are making linear projections based on very thin data. As the article makes clear, enthusiasts tend to equate processing power with intelligence in a linear correlation: the faster our processors, the smarter our machines will become as a natural law.
But software--the ultimate source of "intelligence"--doesn't work like that. Faster CPU cycles don't magically create smarter software. As for the expectation that software will "learn" to program hyper-intelligent versions of V1.01 software, where is the real-world evidence to support this assumption?
Machine learning is highly proscribed. In limited fields, machines can "learn" to achieve better results by cycling through data and correlating data to responses and outcomes.
But as the dashed hopes of general-intelligence AI researchers have revealed, expanding these very limited forms of learning to generalized intelligence is not simply a matter of extending present capabilities via faster CPUs and cheaper memory.
There is also a financial aspect that techno-enthusiasts tend to ignore in their confidence that lower prices for processors and memory are all that matters. They assume cheap AI will mean ubiquitous AI.
History suggests that technology will be employed wherever it reduces costs and increases productivity, i.e. increases profits, and where it creates new markets for the things that humans crave: novelty and convenience.
The idea that there could be limits on AI's leverage on profits, novelty and convenience is heresy to AI enthusiasts. The possibility that the pursuit of profits, novelty and convenience could run aground on our genetic programming or the need to pare away non-essentials (if not for our physical survival, then for our psychological/ spiritual well-being) is also heresy.
Will Super-Intelligent AI Take Over the World? Perhaps it's not just a matter of technological capabilities, but of what is scarce and what is absolutely essential.
What will we scarce and what will be essential going forward? Perhaps not what techno-fantasies expect.
Summary of the Blog This Past Week
Could Inflation Break the Back of the Status Quo? 8/5/16
What Happens When Rampant Asset Inflation Ends 8/4/16
Revealing the Real Rate of Inflation Would Crash the System 8/3/16
Inflation Hidden in Plain Sight 8/2/16
The Burrito Index: Consumer Prices Have Soared 160% Since 2001 8/1/16
Best Thing That Happened To Me This Week
Recorded more of the "Kitchen Sink Concerto" with collaborator/composer C.C.
Market Musings: Tightening Bollinger Bands = Big Move Coming
Look at the tightened Bollinger bands in the daily chart of the S&P 500. This presages a big move up or down in the weeks ahead.
Many technicians have targets above the current level of 2,182 of 2,200, 2,250 and 2,300.
The unstoppable rise in US stocks suggests that the "big move" will be a new run higher. But as the chart shows, breakdowns and breakouts can both be false.
The bullish MACD and stochastics suggest the run higher will continue to at least 2,200, with the potential to leap up to 2,250 or even 2,300.
Will this move higher be a false breakout or the real deal? Only time will tell, but the potential for a false breakout that instills a dangerous complacency should not be dismissed.
Rampant confidence in central banks and complacency about the unstoppable rally are setting up the potential for a "surprising" drop back down to 1,875 to fill the big open gap that beckons.
From Left Field
The climate crisis is already here – but no one’s telling us
100 Most Scenic Restaurants in America for 2016
Meet the China ‘whisperers’ who get the big deals done in Silicon Valley
An Edible Marijuana Pioneer Is Ready for Billion-Dollar Cannabis Brands -- it's about wellness product lines....
I Know Why Poor Whites Chant Trump, Trump, Trump (via Maoxian)
Chinese Furniture Fashion Ravages West Africa’s Savannas -- but growth is always good, right?...
The Humbling of American Tech Giants in China (via Maoxian) -- a protective, intrusive state and thousands of hungry entrepreneurs with the tools and drive to copy and adapt any idea/product....
Meet The Woman Behind China's Largest Ride-Sharing Service (via Maoxian)
War with China: Thinking Through the Unthinkable
Unless Both U.S. and Chinese Political Leaders Decline to Carry Out Counterforce Strategies, the Ability of Either State to Control the Ensuing Conflict Would Be Greatly Impaired.
The Republican Milgram Experiment -- one button brings shocks, the other one a sweet treat...
Harnessing the Immune System to Fight Cancer: New drugs and methods of altering a patient’s own immune cells are helping some cancer patients —but not all — even when standard treatments fail.
We’re in a Low-Growth World. How Did We Get Here? (via Joel M.) -- stagnant demand, prospects and incomes...
"You have power over your mind - not outside events. Realize this, and you will find strength." Marcus Aurelius
Thanks for reading--
charles |
|
|
|
|
|
