More quickly, accurately and less “human”. The future of work in the age of Artificial Intelligence
August 9, 2017
With kindly permission by the Author, we publish an excerpt from the book Humans Need Not Apply. A guide to Wealth and and Work in the Age of Artificial Intelligence by Jerry Kaplan. The Italian translation of the volume has been published by LUISS University Press with the title Le persone non servono. Lavoro e ricchezza nell’epoca dell’intelligenza artificiale
In a nutshell, after fifty years of effort and billions spent on research, we’re cracking the code on artificial intelligence. It turns out that it’s not the same as human intelligence, or at least it looks that way right now. But that doesn’t matter. In the words of computer scientist Edsger Dijkstra, “The question of whether machines can think is about as relevant as the question of whether submarines can swim.” Whether the website that finds you a date or the robot that cuts your grass will do it the same way you do doesn’t matter. It will get the job done more quickly, accurately, and at a lower cost than you possibly can.
Recent advances in robotics, perception, and machine learning, propelled by accelerating improvements in computer technology, are enabling a new generation of systems that rival or exceed human capabilities. These developments are likely to usher in a new age of unprecedented prosperity and leisure, but the transition may be protracted and brutal. Without adjustments to our economic system and regulatory policies, we may be in for an extended period of social turmoil. The warning signs are everywhere. The two great scourges of the modern developed world—persistent unemployment and increasing income inequality—plague our society even as our economy continues to grow. If these are left unchecked, we may witness the spectacle of widespread poverty against a backdrop of escalating comfort and wealth. The work in artificial intelligence is advancing on two fronts.
New systems of the first class, many of which are already deployed, learn from experience. But unlike humans, who are limited in the scope and scale of experiences they can absorb, these systems can scrutinize mountains of instructive examples at blinding speeds. They are capable of comprehending not only the visual, auditory, and written information familiar to us but also the more exotic forms of data that stream through computers and networks. Imagine how smart you would be if you could see through thousands of eyes, hear distant sounds, and read every word as it is published. Then slow the world down to a pace where you can sample and ponder all of this at your leisure, and you’ll get an idea of how these systems experience their environment.
As we amass data from an expanding array of sensors that monitor aspects of the physical world—air quality, traffic flow, ocean wave heights—as well as our own electronic footprints such as ticket sales, online searches, blog posts, and credit card transactions, these systems can glean patterns and grasp insights inaccessible to the human mind. You might reasonably describe them as exhibiting superhuman intelligence, but that’s misleading—at least for the foreseeable future—because these machines aren’t conscious, self-reflective, and don’t exhibit any hint of independent aspirations or personal desires. In other words, they don’t have minds,as we commonly understand the word. They are incredibly good at specific tasks, but we don’t fully understand how they do what they do. In most cases, that’s because there is literally no explanation that can be comprehended by simple creatures like us. This area of research doesn’t have a universally accepted name. Depending on the focus and approach, researchers call it machine learning, neural networks, big data, cognitive systems, or genetic algorithms, among others. I will simply refer generically to the product of their efforts as synthetic intellects.
Synthetic intellects are not programmed in the conventional sense. You cobble them together from a growing collection of tools and modules, establish a goal, point them to a trove of examples, and set them loose. Where they wind up is unpredictable and not under their creator’s control. Synthetic intellects will soon know more about you than your mother does, be able to predict your behavior better than you can, and warn you of dangers you can’t even perceive. I will describe in some detail how synthetic intellects work and why they transcend our common preconceptions of what computers can do.
The second class of new systems arises from the marriage of sensors and actuators. They can see, hear, feel, and interact with their surroundings. When they’re bundled together, you can recognize these systems as “robots,” but putting them into a single physical package is not essential. In fact, in most cases it’s undesirable. The sensors may be sprinkled throughout an environment, on the tops of streetlights or in other people’s smartphones, with their observations harvested and siloed in some distant server farm, which then uses this information to formulate a plan.
The plan may be executed directly, by controlling remote devices, or indirectly, for example, by coaxing you to take some desired action. Often, the results of these actions are immediately sensed, leading to continuous revision of the plan, just as you do when you guide your hand to pick up an object. You are part of such a system when you follow automated driving directions. The program, monitoring your location and speed (usually by GPS), directs you, often pooling your information with that of other drivers to detect traffic conditions, which it uses in turn to route you (and them) more efficiently.
Perhaps the most remarkable of these systems will appear deceptively simple, because they accomplish physical tasks that people consider routine. While they lack common sense and general intelligence, they can tirelessly perform an astonishing range of chores in chaotic, dynamic environments. To date, automation has mostly meant special-purpose machines relegated to performing repetitive, single tasks on factory floors, where the environment is designed around them. In contrast, these new systems will be out and about, tending fields, painting houses, cleaning sidewalks, washing and folding laundry. They may be working in concert with human workers to lay pipes, harvest crops, and build houses, or they may be deployed independently in dangerous or inaccessible places to fight fires, inspect bridges, mine the seabed, and fight wars. (we could say forged laborers).
Of course, these two types of systems—synthetic intellects and forged laborers—can work in unison to perform physical tasks that require a high level of knowledge and skill, such as fixing cars, performing surgery, and cooking gourmet meals. In principle, all these developments will not only free you from drudgery but make you more efficient and effective, if you’re lucky enough to be able to afford them. Bespoke electronic agents may promote your personal interests, represent you in negotiations, and teach you calculus—but not all such systems will be working on your behalf. Humans are suckers for the quick win.