Reproducing Robots

About 100 years ago the word robot was coined. Its origin is a Czech word, robota, which means “forced labor.” In the ensuing century, the definition of the term robot has evolved (thanks to the use of computers) from a machine that behaves as a slave, to an entity that can act autonomously and intelligently by performing complex tasks. In fact, with the advent of artificial intelligence, some people now speculate that we may be on the verge of creating robots that are far smarter than humans, and may soon turn the tables and enslave us!

Robots running amok is a fear that, by dint of unleashed technology, we may about to uncork a machine-like Pandora's Box, by taking an unwise step that will lead to future ominous problems. That possibility has created countless science fiction tales which suggest that our human future may contain many battles with robots who rebel and try to defeat us. One of the most famous depictions was a series of novels by Isaac Asimov, who expanded upon the Czech origin of robot, to coin the term robotics in 1941. Asimov conceived of his three laws of robotics—that would prohibit robots from harming human beings. His assurances, however, have not mitigated the fear in some people’s minds that robots may eventually rebel and some day take over.

Now comes the possibility from a scientific collaboration between teams from the UK and Netherlands, that robots may soon breed and evolve. That idea takes these machines ever closer to human-like capabilities—maybe even enclosing them within the realm of what we call life. Yikes! That's too close to Frankenstein's creation of a “living being” from dead body parts... and we know the troubles caused by that creature.

So can a robot ever attain the qualities of evolving and being “alive”? Biology defines life as exhibiting about six features: (1) a complex assembly of “living” cells, (2) able to respond to diverse stimuli, (3) reproduction, (4) self-regulation, to respond to its environment, (5) possessing homeostasis—to maintain critical internal conditions and balances, and (6) processing energy, to carry on its metabolic activities.

Maybe those 100-year-old robots were simple programmed machines, but today's robots feature most of these characteristics of life. For example, the UK and Dutch researchers' robotic technology is able to spawn (birth?) offspring that receive a transplanted “brain” (in the form of software) from their “parents,” which contain inherited traits.

These robots employ artificial intelligence, which guides that inheritance, so as to optimize certain capabilities for specific missions that we are as yet unable to comprehend. Is this some form of robotic evolution—an evolution in which they can control themselves, and pass traits on to future robotic generations? Are we opening a Pandora's Box, from which super robots will soon breed and become god-like beings? Asimov, we may need your robotic laws to protect us.

AI's Outlook—Part 1

Artificial intelligence (AI) is upon us. From those primitive programming of computer decision-making capabilities just a few decades ago, we now have software that can beat humans at their most intellectually- and creatively-demanding tasks. Recent programs have shamed human champions at chess, Jeopardy!, and even the complex and ancient Eastern game of Go. Other recent spectacular feats of AI include sophisticated robots and self-driving autonomous vehicles.

These incredibly fast and impressive developments have some people ecstatic at the possibilities of future AI applications and other people frightened at what these smart machines may do. What happens when an AI robot becomes far smarter and stronger than humans? Do we need to fear what they might do to us? So far, AI has proven to be superior to the human brain only at narrowly-defined tasks (such as chess and Go), but what will the future bring?

I have been taking an online course from the Delft University of Technology in the Netherlands. One of their researchers did a fine job of scoping out the future of AI. It's pretty complex and replete with differing interpretations, but here's my try at summarizing what he said.

When considering the future of AI, it helps to understand it by looking at three different issues: (1) autonomy, (2) super intelligence, and (3) consciousness.

1. Autonomy is the ability of robots and AI to do things on their own, without human oversight. That said, autonomy is not a crucial issue on its own, since even the thermostat that regulates your household temperature does it on its own. This is important—it's not the autonomy that's the key, it's what the robot will do, and what control might we have over its actions.

Specifically, the concern about the autonomy of AI comes in when it deals with ethical dilemmas. Will the robot's actions be commensurate with human moral values or not? This question comes into focus when we ponder the choice that a self-driving car would make in an emergency. Will it, for example, decide to prioritize its passengers' welfare or that of pedestrians that are in the vehicle's path? How should the AI software program be designed to appropriately reflect human values? Do we even know what that means?

Self-driving vehicles certainly will drastically reduce the many thousands of highway deaths each year due to driver error or inattentiveness. There's no question about that benefit. But what about that one accident in which an autonomous vehicle made a “bad choice” in our eyes? We've already seen a couple of instances; for example, when a self-driving car killed a pedestrian in Arizona last March.

How do we program AI vehicles? We're a long ways away from knowing how. And who is liable, when a self-driving car causes a death? The vehicle manufacturer? The AI programmer?

This concern about autonomy is significantly greater for AI that is used in military robots and drones. Can an autonomous robot, sent in to a dangerous situation (in which the threat to the lives of military personnel might be avoided), appropriately discern the difference between a crouching enemy with guns and a frightened family that contains several children?

More on AI next time...

Smart Machines—Part 2

Something similar has happened in the field of embodied cognition. People have built robots for a long time, and some of them—especially when they look like a human and sort of act like one—have been very impressive. But again, these robots could do only what their designers had programmed their “brains” to do. 

What was frustrating and limiting was that every action the robot did required enormous computing power—yet those actions were quite simple. If the robot encountered something the scientists had not thought to include in its software, the robot would spectacularly fail. Maybe its designers had cleverly (and very complexly) programmed the robot to walk upstairs, but if it stepped on a marble, it'd tumble over and lie incapacitated. And any robot that accomplished impressive feats required such brainpower that it gulped large quantities of energy and drained its batteries quickly. (Our human brain is an energy hog.)

When researchers realized the finesse, efficiency, and proficiency of EC and began to build robots based on this principle, those robots are taking the next quantum leap into the future. The EC principle allowed them to create robots that move uncannily like humans and other animals, without much computer power and without the need for the robot's “brain” to plan and execute every move.

To watch a video of a robot programmed with EC is rather astonishing—if not also a little eerie. (This link gives you a number of such videos: https://www.youtube.com/user/BostonDynamics). Boston Dynamics has built several of them and has demonstrated how capable and autonomous they are. Take one of these robots outdoors, where it encounters terrain that it's never negotiated before—uneven ground, snow and ice, deep mud—and it does better than a human can. You watch one of the robot's feet slip or get bogged down, and it stumbles, awkwardly pirouettes, but quickly recovers. Another video shows a human researcher sneaking up behind a robot and giving it a violent shove with a stick. The robot stumbles forward, catches its balance, and continues its previous activity.

These amazing accomplishments have researchers very excited and furiously engaged in experiments to improve what these smart machines can do. The future in the fields of AI and EC is nearly upon us. What comes next? What novel accomplishments will we soon see? The promises are both thrilling and sobering—even rather ominous.

For example, what happens when the cognitive abilities of an AI computer exceed those of the human brain? It's only a matter of time. They have already demonstrated the ability to best humans in a few kinds of intelligence tests, but yet still fall quite short of the human brain's overall flexible abilities. That threshold will soon be crossed, however. When it happens, will an AI computer then be able to reason, become self-aware, or even possess consciousness?

These questions interfere with the sleep of some scientists and philosophers. They are worrisome to many people. A few people even fear that computers might take over the world and force us feeble humans to be their slaves. This fear has spawned a few fascinating books and movies. Nobody knows what will happen. What is disconcerting, at the least, is that research is moving quickly forward, with little consideration of where we are going or what precautions should be made. (That's an old story with human technology.)

Something similar might be said about EC robots. Most of the current research in this area is being sponsored by the military. What are their plans for these smart machines? Obviously, these robots will someday perform much more effectively on the battlefield than human soldiers—being stronger, faster, and more invincible. The death of a robot—no matter its price—is far less onerous than the death of a soldier. But what might happen when a platoon of EC robots invades what is believed to be a fortified bunker of enemy soldiers, and finds it instead occupied by a group of cowering women and children? Will these smart machines also have the moral sense to halt their invasion?

Both AI and EC robots promise some wonderful benefits. But like so much technology of the past, what was once seen as a blessing sometimes had a dark side. Are we being careful enough?

Smart Machines—Part 1

There are some amazing advances currently being made in two related scientific fields—both of which are exploring ways to create machines that mimic (and even surpass) human cognitive and physical capabilities. In each area researchers have built robots that exhibit stunning skills. Engineers and scientists have tried for several decades to manufacture robots and computers that are as proficient as human beings, with very little success. In the last few years breakthroughs have occurred.

The two similar fields are artificial intelligence (AI) and embodied cognition (EC). The holy grail for many years in AI has been to create a computer that has cognitive abilities equal to that of the human mind. Computers can process data far faster than the human brain can, but the problem has been figuring out how to program a computer to be able to preform the nimble, parallel processing that the brain does; which allows it to instantly recognize faces, or exhibit impressive learning capabilities. That problem is now being solved.

The second smart machine accomplishment has seen the construction of robots that feature embodied cognition. This breakthrough had to wait until scientists could fully appreciate what EC is. In humans (and most animals) most of the things that we do—walk, breathe, swim, or just move the body—are accomplished without any conscious effort on our part.

Do you consciously think and guide your body through every move, when you walk into the kitchen to get a drink of water? What is required to do so is a long and complex string of interacting movements and nerve signals that pretty much do the job on their own. Otherwise, we'd literally be moving in ultra-slow motion, unable to accomplish but one or two activities in a day's time. What allows us to do so many things and to do them fluidly and quickly is EC—wherein the major part of our brain unconsciously engages in myriad activities, so as to free up our thinking brain to ponder more esoteric things like yesterday's events and planning tomorrow's.

The recent AI computers use what is referred to as “deep learning,” where the computer teaches itself. Earlier attempts at AI used the capabilities of super computers to make lightning-fast computations using incredibly complex software programs, but these machines could only do what the human programmers decided to program into them. If the AI computer encountered a novel situation, it was stumped, because nobody had thought to program that particular scenario.

The algorithms being used in the latest AI computers are far less complex (because they do not need to cover every scenario imaginable), but are fundamentally more flexible—like the human brain. The computer teaches itself. Give it the simple rules of chess or Go or Jeopardy!, and it will teach itself by running millions of practice sessions. It even invents novel moves in these games that a human had never thought of. World champions of these three games have recently been humiliated by AI machines.

More smart machines next time...