“A Place to Stand”, by Mitchell Howe Monday, Jan 15 2007
friendly ai and singularity 5:49 pm
Starting in the 1490’s and stretching on for more than 300 years, intrepid explorers searched in vain for a fabled Northwest Passage that would greatly shorten sailing time between Europe and Asia. The hazardous process was costly in terms of lives and equipment, and was exceptionally tedious; vague maps were painstakingly fleshed out with the details of each expedition’s trek into the vast maze of islands and ice that comprise the northern extremes of the western hemisphere. When proof of an actual passage finally became evident, the route was so far north that it was not until 1969 that the icebreaking tanker SS Manhattan became the first vessel to navigate it.
Interestingly, this first navigation of the Northwest Passage came more than eight years after Yuri Gagarin became the first man in space – a testament to three centuries of nautical futility. Had one of the early passage-seekers taken a 108 minute orbital flight like Gagarin’s, it might have been obvious that no practical seaway existed. At the same time, passing over Central America could have produced the idea for an ocean-linking canal across the narrowest stretch. Indeed, it might have seemed possible to stretch out a hand and scratch out a canal as simply as wiping a smudge off the window.
Perspective is a powerful thing. It reveals the elegant solutions to complicated problems. It also provides an advantaged position from which to act on these solutions – an important consideration, since the most powerful tool is often worthless unless it can be effectively applied. Archimedes, the great mathematician of ancient Greece, famously described this principle in his discussion of a simple tool: the lever. “Give me a place to stand,” he said, “and I will move the earth.”
Unfortunately, the gravest medical conditions and social problems today cannot be solved by something as simple as a huge lever in earth orbit. However, the leverage principle still applies. It is our ethical imperative as a global family to improve the human condition, but we owe it to ourselves – and especially to those who suffer most – to find the most effective and efficient means possible.
In a very real sense, we are using 21st century technology to map the contours of insidious diseases in a manner befitting the 16th century. Genome databases, for example, prevent much duplication of effort, but human eyes gain surprisingly little insight from examining immensely long strings of code that did not evolve with readability in mind. Molecular biology remains, to a large extent, a trial-and-error process.
Many social and economic problems are similarly stubborn, despite a seeming abundance of information about them. Sadly, the last century has not seen child abuse and poverty rates fall anywhere near as quickly as the prices of consumer electronics. These trends in science and society say as much about our own innate abilities as they do about the problems. Ironically, human intelligence falls short of being able to make immediate sense of the genetic blueprints behind it or the civilization proceeding from it.
Though lacking in many ways, our brains perform magnificently in others. Perhaps our greatest talent is making rapid conclusions based on visual information. It is no small feat to immediately recognize a friend’s face from different angles, in different lighting conditions, with different possible expressions – yet we succeed at such tasks routinely and consistently. In fact, our visual processing circuitry is so advanced that it has evolved to play important roles in replaying memories and working out problems. For this reason, we often have trouble understanding concepts we cannot “visualize.” Also, just as our eyes tend to focus on only one thing at a time, we have difficulty with problems that involve many interacting components – stubborn scientific and social issues being perfect examples. Computer systems now assist experts in many fields, but writing software of such complexity is itself a tremendous challenge, and even when these programs make billions of calculations we could never have time for, they don’t visualize the problems – or the solutions – any more than a convention of early explorers could hope to visualize a northwest passage by swapping adventure stories over cocktails.
One of the greatest promises of artificial intelligence (AI) is the possibility of minds not only of equal or greater intelligence to our own, but minds capable of reconfiguring themselves to solve problems that are very different – yet perhaps no more complex – than making sense of human-style visual input. Such minds may be able to “visualize” molecular biology or socioeconomics as effortlessly as we would visualize the characters of a favorite novel. AI could also be configured to readily conceptualize computer programming code, facilitating improvements to its own design. And, most importantly, AI could share and expound on our human understanding of ethics and compassion – a trait called Friendliness by some researchers. As a benevolent partner with unprecedented perspective on a huge variety of problems, Friendly artificial intelligence could improve our lives from a place of enormous leverage.
This place – the moment when greater intelligence makes its historic debut – has a name: the Singularity. We can’t know ahead of time what the first artificial minds will see from their stratospheric heights of understanding, but we can be confident that tools will be perfected to act on any elegant solutions found. Likely candidates are already under development, such as nanotechnology, which holds the promise of near-infinite control over our environment through the engineering of tools at the molecular level. Nevertheless, it makes sense to invest in Friendly AI first. Not only is AI more likely to be feasible than advanced nanotechnology in the near term, it will facilitate success in all other endeavors. In fact, from the Singularity’s vantage point, our current paths of technological progress may well resemble misguided journeys to the frozen north. Perhaps we are packing parkas and biscuits when what we really want are mosquito netting and steam shovels.
It’s a tribute to the human spirit that we seek to uplift those around us and move society ever forward. It’s a tribute to human ingenuity that we continue to develop better tools to overcome what we see as the greatest obstacles in this endeavor. But in light of our inherent difficulty in solving many kinds of problems, it is appropriate to seek helping hands and fresh perspectives. People are suffering, and greater leverage can help.
If you give a man a fish, you feed him for a day. If you teach a man to fish, you feed him for a lifetime. But if you discovered a simple way to make food optional, you, the man, and the fish could move on to other pursuits. What are we missing? With Friendly AI on our side, we’ll have the perspective needed to find out. The Singularity is more than just a lever long enough to move our world for the better. It is a place to stand.
©2002 Mitchell Howe
