Ray Kurzweil is talking about being able to predict the advance of information technology. The key to success of being an inventor is in being able to estimate when technologies will be advanced enough to make an invention possible – timing is critical. Over thirty years of doing this he’s found that the models he worked out to predict future technology trends has worked pretty well.
If we can measure information content of an area, we find that growth is exponential, with roughly a doubling every year – the power of technologies expands by a billion every 25 years, while the size of technology shrinks by a factor of a hundred every ten years.
Ray shows a new pocket-sized print-to-speech reading machine – he’s been involved with reading machine technology for thirty years now. In 2002 he predicted that technology would be available for a pocket reading machine would be available in May of 2006, and that development of such a machine would take roughly four years. In July they introduced the pocket reading machine.
Ray goes on to talk about the extent to which artificial intelligence programs are providing generally useful functions now, and the narrowness of these applications are getting less narrow over time.
The Paradigm Shift Rate is now doubling every decade. People tend to extrapolate in a linear way, when progress is almost always faster. The rate is actually accelerating. He gives examples – the phone took half a century to be adopted by half of the US population – the cell phone did it in seven years. Another trend is the democratization of knowledge creation, which is fueling an explosion of knowledge. The amount of knowledge is also growing exponentially, doubling every year.
He shows a graph that he calls Countdown to Singularity, which shows technological evolution in a continuum with technological evolution, in a straight logarithmic trend line.
People criticize Kurzweil for thinking that exponential growth can continue – and it’s true that each particular technology runs out of steam, but new technologies evolve that keep the trend going. When Moore’s Law runs out in current chip design, we’ll see three dimensional molecular computing rise to keep the growth going. There’s nanotube-based memory set to hit the market next year.
By 2013 we should have computers that can equal the processing power of all regions of the human brain.
It’s remarkable how smooth the trend lines are, given the vagaries of the activities of millions of people. It’s like other examples in science where we see predictable activity come out of random, chaotic individual events. The classic example from the 19th century is thermodynamics. Technology evolution just such a case of predictable behavior.
Our consumption of information technology more than keeps pace with the growth of capacity.
In biotechnology we’re using technology to reprogram biology. He’s involved with one company that has cured pulmonary hypertension in animals by injecting a new gene – it’s now going into human testing. There are thousands of these developments happening now. This is a new paradigm in drug development, designing drugs using technology instead of just discovering them.
He talks about repirocytes – robotic red blood cells that are already being tested by animals.
The ultimate source of utilizing the power of information technology will come from reverse engineering the brain – we’re now getting to the point where we can scan brain data to see individual activities. The design information of the brain is a billion times simpler than the apparent complexity of the brain – we know this because of the amount of information the genome can contain. We will succeed in modeling the brain within the next twenty years – which will fuel truly intelligent systems.
Ray shows a video of a prototype of a translating telephone, where he speaks in English and the person on the other end hears him in German, and vice versa (also in French). He says these systems will be common in cell phones in the next ten years. This translation is done through pattern recognition informed by large databases.
By 2010, computers will begin to disappear – images written directly to our retinas with ubiquitous high bandwidth connections at all times and electronics so tiny they’re embedded in clothing and glasses. Full immersion virtual reality will be feasible, and augmented reality (eyeglasses that tell you the name and birthday of the person you’re talking to) and effective language translation will exist.
By 2029 we’ll have 30 doublings. $1k of computation will buy 1,000 times the capacity of the human brain, reverse engineering of the human brain will be completed, computers will pass the Turing test, and nanobots will provide expansion of human intelligence.
Human life expectancy was in the 20s when life evolved. By 1800 it had reached 37, by 1900 it had reached 48. According to his models in twenty years we’ll be adding more than a year of life expectancy to the lives of living people every year.