“The Greatest Leap”, by Mitchell Howe Tuesday, Jan 23 2007
friendly ai and singularity 7:24 pm
On a blustery December day in 1903, a pair of bicycle mechanics made good on a challenge mankind had given itself since it first looked up into the sky and envied the perspective and freedom of the birds. These two men, brothers, recognized that the thousands of years of technological progress on which they stood were sufficient to enable a glorious leap into a new era: the age of powered flight. Having diligently combined the latest advances in gliders and gasoline engines with their own experimental knowledge of aerodynamics, Orville and Wilbur Wright took their prototype out to the dunes of Kitty Hawk and made history. Over the next hundred years our world shrank and our frontiers expanded heavenward, thanks to two men with a keen sense of the times in which they lived and a compelling vision of what was possible.
For those who take the time to follow trends and reflect on the dizzying pace of technological innovation, a similar revelation can present itself; a conclusion so momentous that, as was the case with those who initially found the idea of controlled flight absurd, it can be somewhat hard to swallow. Progress has reached the point where another historic leap is possible, one arguably more significant than all the advancements of the last thousand years put together.
From the first primitive flints and spears, civilization has moved forward by creating tools to improve the quality of life, and on using these tools to create still better tools. Every modern miracle, from antibiotics to the international space station, owes its existence to our having successfully built upon the achievements and discoveries of our predecessors. But there is one tool that we have never managed to improve upon, even though it must ultimately take credit for every social, spiritual, or technological leap we make: our intelligence.
The groundwork at last exists for a project that will culminate with the creation of greater-than-human intelligence. In fact, a number of avenues are now open to achieve this, including genetic modifications and direct brain-computer interfaces. But the approach most likely to be safe, predictable, and practical in the near term is the creation of true Artificial Intelligence (AI). AI, if designed correctly and responsibly, will be able both to improve its own capabilities and to share our human understanding of morality - an ability called Friendliness by some researchers. Thus, we can expect the earliest Friendly AI to quickly become an invaluable member of the human family, vastly increasing our collective ability to solve problems and resolve differences.
This creation of greater-than-human intelligence, a milestone known as the Singularity, will explode our current models of technological progress, removing one of the limiting constants from our equations describing change. Infused with the creative equivalent of rocket fuel, exponential growth (the current trend in many industries) will become hyperexponential growth, with advances coming so quickly that the world as we know it will be abruptly discontinued. In its place will be a reality so brilliantly crafted that we may be incapable of imagining it with our current level of intelligence.
What we can imagine is this: today’s discontents giving way to the more dignified pursuits of mankind, with the galaxy itself opened to our exploration and discovery. Nanotechnology - the engineering of materials and machines at the molecular level - will permit universal material prosperity, through self-replicating factories that produce goods of unprecedented quality at negligible cost. Combined with genetic engineering, this will mean the end of disease and aging. Such capabilities are readily foreseen given that we have already made important strides in both these fields; greater intelligence will allow us to master them, and to rapidly deliver on their enormous potential.
Since the rise of civilization we have seen no appreciable increase in our level of intelligence. Remarkable as our human intellect may be, it is still a flat line along which we have shuffled in pedestrian fashion for the last 50,000 years. But we need not confine ourselves to this dusty old road. We can fly.
For above this flat surface of intelligence - above the mindset of daily struggle and business as usual - is a larger view of history and reality: the perspective of a fully empowered society that defines poverty as having to work to make ends meet, that remembers involuntary death as a disease that was cured in the early 21st century, and that knows scarcity only through historical accounts.
But no matter how many pieces we have lying around, we will not make the leap to greater intelligence - to the Singularity - unless we make the commitment to put it all together… and to do it right. AI, especially Friendly AI, simply won’t design itself (at least not the first time!). The creation of greater-than-human intelligence is the greatest project ever to be undertaken by our species, but is entirely doable within the next 10 to 20 years - or sooner - if enough people are willing to get involved. There is no reason to wait. Let’s get out to the dunes and make history.
©2002 Mitchell Howe

January 23rd, 2007 at 8:09 pm
BRAVO!!
January 24th, 2007 at 2:02 pm
Perhaps it is time for a Manhattan Project devoted to developing Friendly AI.
January 24th, 2007 at 6:22 pm
“Perhaps it is time for a Manhattan Project devoted to developing Friendly AI. ”
We’re working on it, but we still need the Manhattan Project’s prestige, resources, and billion-dollar budget. Do we know any billionaires who are interested in helping us?
January 25th, 2007 at 5:33 am
A large government sponsored project might be one approach to the problem, but this approach does not have a particularly good track record. Money alone cannot solve this problem. I suspect that it’s going down to the bicycle engineers - a lone individual or small group - who will come up with the first workable AI design.
January 25th, 2007 at 1:22 pm
someone contact bill gates.
January 26th, 2007 at 8:52 am
I’m skeptical that AI will cause the firestorm of progress that people claim it will, at least in the short term.
I don’t have a background in R&D but it seems to me that much of the work involved doesn’t revolve around flashes of brilliance but a steady slog of trial and error. A lot of our best scientists don’t even have exceptionaly high IQs. And conversely the top two IQs in the USA belong to complete societal rejects.
I don’t doubt that a friendly smarter than human AI will prove useful I’m just skeptical that it will drive progress that is currently imagined.
January 26th, 2007 at 1:51 pm
“I’m skeptical that AI will cause the firestorm of progress that people claim it will, at least in the short term.”
You’re thinking of AI as being an exceptionally smart human, which is roughly as inaccurate as saying that a human is just an exceptionally smart amoeba. Imagine what human civilization would seem like to an amoeba (if amoebas could think); more to the point, imagine how amoeba philosophers denounce these “humans” as being incapable of doing anything radically new, as history shows that even exceptionally smart amoebas don’t fundamentally change the amoeba way of life.