The Works and Ideas of Futurist Ray Kurzweil

There’s a moment near the end of “Transcendent Man,” a new documentary about futurist Ray Kurzweil.  The camera follows Kurzweil to the lip of a seashore as he silently watches the waves churning in and out.  “What are you thinking about?” an off-camera voice asks.  “I’m thinking about computation,” he says, smiling.

Ray Kurzweil is not an easy man to understand.

And neither are his ideas regarding the future, most fully articulated in his latest book “The Singularity Is Near.” Long an innovator in the field of computer science and pattern recognition, Kurzweil has turned his attention to genetics, nano-technology and artificial intelligence and asks, ‘What comes next?’

His answers can be dazzling…and a little unnerving.  Just a few:

• “We’ll reprogram our genetics away from disease and aging.”
• “We will have blood cell-sized devices that go in your brain and allow us to merge with non-biological intelligence.”
• “In about 20 years  a computer will be able to match human intelligence and surpass it.”
• “We will merge with these machines be able to download our brains, and in effect live forever.”

All this and more…by the year 2029.  Welcome to the Singularity.

The key to understanding these and many of his other predictions is the concept of exponential growth and what Kurzweil calls the “Law of Accelerating Returns”:

“The nature of technological progress is exponential.  If I count 30 steps linearly – 1, 2, 3, 4, 5 – I get to 30.  If I count exponentially – 2, 4, 8, 16 – 30 steps later, I’m at a billion.  It makes a dramatic difference.”

Exponential growth, he says, yields extremely powerful results in shorter and shorter periods of time.  Examples are all around us:

“The quintessential example are computers…generally in the case of computers they double in power for the same cost in less than a year.   When I was a student we shared a computer that cost tens of millions of dollars and took up a building.  The computer I have on my belt now is a million times cheaper, and a thousand times more powerful.  That’s a billion-fold increase in price performance.  So we’ll do that again in the next 25 years.”

Kurzweil’s predictions may sound fantastic, but he has an uncanny track record.  Decades ago, he predicted the fall of the Soviet Union, the unlocking of the human genome, and the year a computer would beat a human at chess (he was right on all three counts.)

Now he’s predicting that rapid advancements in brain science will yield exponential advancements in artificial intelligence.  “The Singularity” is, for Kurzweil, that moment when machines think as well as humans.

“The human brain is not very efficient at computation,” says Kurzweil.  Electronics are millions of times faster.  Once researchers fully understand how the brain works, they’ll be able to build machines using the same principles.  Very quickly, he says, those machine-brains will become much smarter than the human brains that inspired them:

“We’re making exponential gains in understanding the human brain.  There’s actually a fair amount understood already, and we’re getting what we call biologically inspired algorithms – methods that are inspired by how the brain does things.”

We’re understanding how the neo-cortex can take a whole bunch of symbols, call that an idea and give it a symbol, and use that with other symbols to create another idea, and give that a name, and we create a whole hierarchy.”

This has been an exponential progression that’s gone on for the last 125 years.  Anything having do to with information – it experiences that same exponential growth.”

With exponential growth in power comes more and more human-like thought.  While no computer has yet passed the Turing test, they’re getting closer every year.  And just in the last few months, “Watson” – the IBM computer that beat humans on the game-show “Jeopardy!” – demonstrated what Kurzweil terms an “uncanny ability” to understand the subtleties of natural human language:

“It understood the natural language queries – these kind of convoluted brief language expressions that are the answers for Jeopardy! – but then it had all this knowledge.  It knew the kings of Spain in the 15th century and millions of other things.

How did it get that knowledge?  That was not spoon-fed by all these scientists; they didn’t sit there and create a database.  They just dumped in Wikipedia, and encyclopedias, and hundreds of thousands of pages of natural language documents.  Watson read that and understood that and understood the natural language, and created this knowledge base just like a human would.”

Of course, there’s a wide range in human intelligence and knowledge.  Some people are mathematical whizzes, some brilliant composers, and some have the gift of empathy.  But… some are lazy, rob banks, or kick puppies.

“Well, it’s actually kind of dumb to kick a puppy,” says Kurzweil, “and not very smart to rob a bank.  We’re going to try and teach machines the better qualities and more intelligent activities of humans.”  However, he acknowledges, there are no guarantees in the Singularity.

Call them super-computers, smart machines or AIs – Kurzweil’s point is that they’re coming very soon, and that humanity will merge with them.

“We’re going to have robotic white blood cells that are smarter than your white blood cells today.  For example, they’re going to figure out that cancer is an enemy – your white blood cells today think it’s you.  They’ll go inside your brain and put your brain on the cloud of computing.”

2029 isn’t far away, but at present at least it seems that computers have a very long way to go.  Human brains may be slow, but they’re still the only things we know are self-aware.  Computers can drive cars, but they can’t decide where they want to go.  They can diagnose an illness, but can’t comfort a nervous patient.  Computers can’t write jokes…or if they can, they don’t laugh.

This emotional intelligence – sharing a joke, wiping a tear – is exactly what humans do that AIs can’t.  At present.  But Kurzweil says soon, they will:

“This kind of emotional intelligence – being loving, being sexy – these are actually very complicated intelligent behaviors.  And that’s exactly where humans actually still excel.  When I say that computers will pass the Turing test and match human intelligence in all these different ways, that’s exactly what I’m talking about.  Because if we were only talking about logical intelligence, computers are already smarter than we are.”

There are questions that no futurist or technologist can fully answer.  Questions like: will a computer truly experience what it seems to experience?  Or will it be just a fantastic simulation – and how will we be able to tell?

“There’s no objective way to determine subjective reality…but it’s also not an issue we can’t ignore.  We will debate this with machines.  My prediction is that we will accept that they’re conscious because they’re going to be so convincing.   Anyway, they’re going to be very smart, we’re not going to want make them mad at us.  So we’re going to believe them when they say that they’re mad at us or they love us.”

Humanity may not have long to decide.  2029 is getting near.