Sunday, January 1, 2017

Resistance is Futile


This time I look at a book that claims to reveal the shape of our technological future. The technological climate is changing. Ice is turning to water, and there’s no going back.

The Inevitable: Understanding the 12 Technological Forces That Will Shape Our Future by Kevin Kelly (Viking, NY, 2016, 336pp, ISBN 978-0-525-42808-4, $28.00, www.penguin.com)

Kevin Kelly has been on the forefront of the online connected world for more than 30 years. He contributed substantially to the Whole Earth Catalog, the WELL, the Hackers Conference, and Wired Magazine. In his 1994 book, Out of Control: The New Biology of Machines, Social Systems, and the Economic World, he was already exploring some of the themes that appear in The Inevitable. His 2010 book, What Technology Wants, (Micro Review March/April 2011) explores the idea that technologies have characteristics that make them likely to evolve in some directions but unlikely to evolve in others. This is the basis for “inevitable” in his current title. Kelly’s technological forces are what we might call trends, and because he describes them as processes, he assigns a present participle to each. This is a little contrived, as when he uses the participle “cognifying” to describe using artificial intelligence (AI) to make devices or processes more capable.

The trends Kelly describes are in plain sight, but what he reports as already happening on the leading edges of those trends was new to me. He also shows ways in which these trends interact and reinforce each other, magnifying the effect of each. For example, one of his themes is the maturing of virtual reality, which Kelly claims has not become more capable in the last nearly 30 years but has become much cheaper because of technologies developed for mobile phones. High-resolution screens and motion sensors cost a tiny fraction of what they cost in the late 1980s, so inexpensive systems now provide a realistic sense of being somewhere you’re not, while AI and big data enrich the experience.

AI

Since at least as far back as the 1960s, AI has been just around the corner. We’ve turned that corner now, but not in the way many of us feared. Rather than the moody and self-protective HAL 9000 of Arthur C. Clarke’s 2001, we see application-specific bits of intelligence. Driven by our ability to handle big data and our drive to collect and annotate information about everything anyone does, artificial intelligence will be cheap and ubiquitous. As Kelly puts it, “Even a very tiny amount of useful intelligence embedded into an existing process boosts its effectiveness to a whole other level.” He imagines a grid from which you can take as much cheap, reliable AI as you need. Like electricity a century ago, this capability will spawn countless new businesses built on the model of adding AI to some previously unenlightened process. For example, intelligent clothing will communicate with washing machines to control water temperature, the amount of soap, and the intensity of the agitation and spin cycles. More significant than such trivial, but potentially lucrative, applications are AI-enhanced medicine, transportation, weather forecasting, and investing. Such applications exist now and will become more sophisticated tomorrow.

Kelly attributes the success of today’s AI, despite all the false hopes of the past, to the following factors: cheap parallel computation, big data, and better algorithms. Hardware like multi-core microprocessors or massively parallel graphics processing units (GPUs) make possible analytic techniques that would have been prohibitively expensive with older equipment. Huge collections of data about everything from chess games to search results, to tracking cookies make it possible for AI to learn and improve. Hierarchical algorithms, called deep learning, make full use of parallel computation to support such AI successes as IBM’s Watson or Google’s search engine.

Sharing

Our digital culture is communal to a high degree – socialism, but without the state. Wikipedia, Creative Commons permissions, crowd funding, peer-to-peer loans, Tor, Digg, Reddit. Pinterest, and Tumblr are all examples of it. Many people contribute, and everyone consumes without charge. Apache and Linux have unpaid workforces the size of a small town. Over the last hundred years, free markets solved problems that governments could not. Now collaborative social technology is solving problems that the free market cannot. Google, Facebook, and Twitter depend on such collaborative contributions to provide valuable services free of charge, but make huge amounts of money by using AI and big data to deliver targeted advertising. The bottom-up model of user-generated content is a wonderful way for new ideas to evolve and bloom in niches, but as Wikipedia and other examples show, some top-down curation is necessary as collaborative projects grow and mature.

The nature of bits

Several of Kelly’s trends arise from the nature of bits. Because bits are ephemeral, they are easy to gather, duplicate, and rearrange. From this he deduces the inevitability that they will be gathered, duplicated and rearranged, despite attempts to prohibit these actions. This situation stresses our legal and social systems and changes our way of life. Ownership gives way to access. Like the hunter-gatherers we descended from, we’ll soon own nothing but have access to whatever we need. Solid products give way to fluid services that keep updating. A black touch-tone phone on a desk gives way to a continually updated smart phone with much of its data stored elsewhere. People consume music, movies, and “books” online rather than filling their homes with the concrete embodiments of those things. They actively mix, match, and hyperlink fragments of audio, video, text, and images into new creations that befuddle existing copyright laws. Because copies are free, you must make your living by selling trust, immediacy, personalization, discoverability, and other things that can’t be copied. Bits exhibit a network effect. A bit is more valuable if accompanied by metadata – other bits that describe it. A bit is more valuable if linked to related bits. The cloud draws its value from its ability to support and leverage the nature of bits.

Remixing

Duplicating and rearranging bits is called remixing. According to Brian Arthur (Micro Review March/April 2011), all new technologies derive from combinations of older ones. The same goes for combinations of bits. Copying, rearranging, annotating, and linking to text is easy because of our tools. Future tools will enable us to do the same with images and video. Kelly envisions the ability to link from an article on Asian clothing to the fez worn by a character in the movie Casablanca. This will depend not just on new tools but on automated assignment of metadata to every bit of information in the cloud – a small extension of what Google already does. Already, trillions of photos are online, and AI has produced filmable 3D images of many things (for example, the Golden Gate Bridge) from those photos. In a gesture toward recognizing intellectual property and ownership amid all this remixing, Kelly alludes to Jefferson’s distinction between a house and an idea. He proposes to distinguish between copying and transforming and to give free license to the latter, a departure from current copyright law.

Tracking

One form of gathering bits is tracking. Websites, cell phones, social media, and credit cards track our visits and actions, but we also track ourselves. We continually track our exercise, vital signs, and other measurements. This can lead to establishing baselines to support personalized medical treatments. We collect email, record public talks, and may someday, as a few people do now, automatically record all of our interactions. This can give us augmented memories of people, places, conversations, and events – an enhancement of our natural abilities that I’m sure we’ve all wished for at one time or another. In addition to data from tracking ourselves, the internet of things (IOT) gives rise to huge tracking possibilities. Kelly recognizes two models for tracking. In the big brother model, “they” know everything about you. You know very little about them. In the small-town model, tracking is more transparent. You know who’s watching you, and you have a good sense of what they’re planning to do with the information. Bitcoin and public key encryption illustrate the small-town model.

Kelly imagines a slider you can use to control the balance between privacy and openness in your public dealings, and he points out that most people seem to prefer openness. This is because the more you reveal of yourself, the more personally others can treat you. The more private you are, the more generic the services you receive. When it comes to anonymity, the ultimate privacy, Kelly says it’s like a heavy metal – essential to your nutrition, but fatal if you get too much. Everyone who has read anonymous online comments knows this. Privacy depends on trust, which requires a persistent identity.

I find Kelly’s vision of the future of tracking a little ominous. He sees the volume of tracked data growing to the size of an elephant, compared with the mote of dust we track now. This qualitative difference puts it beyond what humans can comprehend – if it isn’t already. He sees all of it reorganized into structures that only machines and AI can work with. They will parse this huge body of information into tiny elements and recombine them in unimagined ways. How we will relate to this planet-sized machine is unclear.

Filtering

Our attention is a scarce commodity. Humans have limited capacity, and there is little we can do about that. According to Kelly, each year brings 8 million new songs, 2 million books, 16,000 films, 30 billion blog posts, and 182 billion tweets. Many filters are available – for example, the Amazon or Netflix recommendation engines, driven by big data and AI – but even if you filter out everything that isn’t perfect for you, you still don’t have time to consume it all. And behind the scenes, a gigantic filtering mechanism matches advertisers with opportunities, trying to show you ads you’re likely to respond to. This is a multi-billion-dollar industry.

Filtering can lead to the kinds of sharp divisions seen in political systems. If different groups see only material that comports with their views, those groups might stop listening to each other and ultimately live in separate realities. Kelly offers no answer to this problem.

Jobs

This is a technology book, not a political one, but as anyone who paid attention to the 2016 US election knows, people care about jobs. Kelly believes that humans should not do anything that machines can do. Rather we should work alongside robots, treating them, a la McLuhan, as extensions of ourselves. He is sure that we will dream up new jobs that we can’t imagine now, just as farmers of the early 1800s could not imagine the jobs of their twenty-first century progeny. This may be true, and perhaps that’s all he needs to say, but it certainly raises questions. For example, can an economic system that efficiently allocates scarce resources adapt to the case in which resources are plentiful and necessary jobs are few?

Questions

Raising questions ties to one of Kelly’s themes. Today, humans ask the internet two trillion questions each year and get good answers – a service that’s valuable but free. That number will grow rapidly as technology enables answers to more personal questions like “Where’s Jenny?” or “When is the next bus?” Kelly thinks search will become an essential universal commodity in the next few years. But answers are not as important as questions. Kelly quotes Picasso as saying in 1964 that computers are useless because they only give answers. Someday computers may be good at asking questions, but for now, this is one of the jobs Kelly reserves for humans. For example, humans will probably always be better at asking, and answering, questions about what humans would like to do with their free time or how they’d like to use each new technology. Kelly makes an analogy between surfing the internet and dreaming. Both feature quick changes of focus and mix the real and unreal. Both blur the distinction between work and play. Both seem like a waste of time, but both can lead to novel juxtapositions of ideas. Ultimately, they can engender questions as profound as Einstein’s asking what you would see if you were travelling on a beam of light. It may be a long time before machines can ask questions like that.

The holos

Kelly’s timeline for most of the inevitabilities he discusses is the next 30 years, but he describes this as the beginning of a century-long process. All his trends merge into one large invention: a new mind for our old species. The new mind has planetary scope and gives us perfect search and total recall. He calls it the holos and defines it as “the collective intelligence of all humans combined with the collective behavior of all machines, plus the intelligence of nature, plus whatever behavior emerges from this whole.” The hardware of the holos already comprises a sextillion transistors, a trillion times the number of neurons in a human brain. And everyone who surfs the web teaches the holos something about what we consider important. By 2025, Kelly estimates, 100% of the planet’s population will have nearly free access to the holos.

Kelly likens these developments to a phase change, like the transition from ice to water. He rejects the term “singularity” in the sense of an exponential growth of AI that makes humans irrelevant. Instead he sees a symbiotic relationship between humans and technology. The details of how it works are unknowable, but the general direction, in his view, is unmistakable. Time will tell whether he is right or wrong, but reading his book is an eye-opening adventure. I recommend it highly.


This article appears in slightly different form in the Jan/Feb 2017 issue of IEEE Micro © 2017 IEEE.