The Electric Pulse

Ray Kurzweil is given a big write-up in Newsweek magazine (here). The article is pretty skeptical, which I guess shouldn’t be surprising, and of course some skepticism is certainly reasonable. My problem with this article is that it dismisses Kurweil’s predictions without posting any alternative. Computers are getting faster, that’s a fact. It’s also a fact that this increase is exponential, and that it has held up now for decades. Maybe you think the rate will slow down- that’s fine, but then you aren’t arguing against Kurweil’s scenario of a Singularity, just his dates. I’d really like one of these MSM types doing profiles of Kurzweil or Aubrey de Grey to get specific at to why their ideas are unreasonable. And besides a civilizational collapse, what possible alternative futures do you see developing?

DARPA (the US military’s experimental research unit) is trying to analyze the basis of thought, and come up with a thermodynamic equation for inteligence. Their goal is to:

demonstrate the first intelligence engineered from first principles

Once again, this is another interesting example of science being far ahead of mainstream views on human and machine intelligence. The project is aimed at reducing all human thought and emotions to equations, with the goal of replicating them in an inorganic substrate. Despite this getting funding today, in 2009, the idea of true artificial intelligence being not so far away is still a cult philosophy.

IBM Machine will compete in Jeopardy

As if the rise in netbooks wasn’t happening fast enough, ARM is planning on introducing a whole line of sub $200 netbooks. The computers will have linux operating systems and battery life more on par with cellphones than laptops.

It’s amazing how quick the cellphone and computer markets have converged. Increasingly, a computer is just a processor that can tie into the internet, making all sorts of new pricing models and form factors possible.

I was just looking at these slides from AMD concerning their new 4770 graphics processor. It is a ~$95 dollar card that computes at ~950 Gigaflops per second, meaning it costs about 10 GFLOP/$1. If Kurzweil and others are right, and the human brain operates at ~10,000,000 GFLOPs, you could now buy the equivalent processing power for $1,000,000.

Of course, if you bought 10,000 of these cards you’d be SOL without the tons of hardware needed to hook them together, and even after doing that they wouldn’t be able to communicate with each other at anywhere near the rate that human neurons do. Still, in terms of pure, raw processing capabilities, we’re at $1,000,000 and falling. Interesting times.

Artificial intelligence has arrived. No surprise- it’s Google that’s out of the gate first with CADIE (Cognitive Autoheuristic Distributed-Intelligence Entity).

You can read Google’s post here, but for the good stuff, I’d skip straight to CADIE’s self built website.

First, there’s the good news that the Chevy Volt is 100% on schedule to be in showrooms late next year. Hoping to provide some competition is the more upscale, and fully electric, Tesla Model S.

The model debuted today in California. The car will cost $57,400 (effectively $49,900 after the $7,500 government rebate), run for 160 miles (or 300 with the premium battery pack), and be ready for sale in 2011.

For one, I’m not sold on the styling. It seems to be a cross between the smooth lines of the Jaguar XK and the boxier look of BMW/Cadillac. I think the end result doesn’t quite blend.

Also, the 160 mile range is fairly problematic. Essentially, to be comfortable, you can only drive about 1 hour in any direction to have enough energy to get back home. That’s pretty limiting. And of course, the range will only get worse as the battery pack ages. It will be interesting to see how much the extended range battery adds to the cost of the car.

And here’s my big thought on the Electric car issue- shouldn’t that $7,500 tax credit spur the development of a CHEAP electric vehicle? I mean, that rebate has a much larger effect on a 20K car than a 60K car. Here’ how I think it could work- without a battery, the cost of producing an electric vehicle is less than a normal one. In other words, you could make a pretty decent electric car with a selling price of $12,500, and then lease the $20,000 cost of the battery. Now you’ve got a car that costs $5,000 up front with a $300/month lease. And since you can fill up for a $1 and should have no maintenance, that car would be a pretty decent deal, not to mention the cool factor. If someone comes up with this in the next year or so, remember- you heard it here first.

We always like new Aubrey de Grey info at the electric pulse. Friend of the blog Michael Graham Richard has a new post with a link to Aubrey’s latest talk as well as an archive of some of his talks in the past. The new talk is very specific and technical so go to the old stuff if you’re new to him.

The Electric Pulse’s interview with Dr. de Grey is here

A club in Barcelona is offering to implant RFID tags into club VIPs This allows them to keep their wallet at home and still buy a round for the house.

Given that this technology has been around for a while now, I’m surprised the major credit card companies haven’t offered this yet. Besides convenience, this would assure that no one could physically steal your card (you could still lift the number and use it online of course). It would also eliminate some of the collateral damage of credit card theft, like when my card stopped working on an East Coast trip because Capital One assumed it had been stolen. And of course, always having the card means you’d use it more, which means more money for the credit companies. Seems like a win-win all around.

…they pull me back in.

Just when I’m about ready to give up reading science magazines with their outlandish claims of what will be possible in the next 12 months, I read the follow up about this company.

Last year, they made the fairly outrageous claim that they’d sequence genomes for $5,000 sometime in 09. Well, apparently, they are going to do just that.

This is just ridiculous. There’s the law of accelerating returns, and then there’s been what has happened in the realm of gene sequencing in just the last two years. In 07, it was big news when sequencing dropped to $1 million dollars. Last year, people were surprised to see it drop to $60,000, and now it looks as though we’ll see another 10+fold drop in price this year as well. Two years ago you were seen as very optimistic if you predicted a $1,000 genome by 2012, and now it looks very possible that we could get there by next year.

Chris Rock has a funny comedy routine about how doctors today never “cure” anything, but just allow you to live with the disease. To a large extent he’s right, just because the diseases that have survived into the modern world are usually so complex that a simple cure is out of reach.

But that tide might be beginning to turn. Doctors recently cured a type of “bubble-boy” disease using gene therapy (article here). Cured, as in the patient before hovered between life and death while receiving mega-doses of antibiotics, and now they live totally normal lives.

This particular disease variant is caused by a single-gene defect, and so is relatively easy to repair, but it won’t be long before more complex diseases die a similar death. It will be interesting to see how the public, which I think (like Chris Rock) is conditioned to see medical cures as incremental, starts to re-act to these new unequivocal cures.

New Scientist reports here on probabilistic computer chips that run 7 times faster and use 1/30th of the power of regular processors. They do this by shifting resources away from keeping every individual calculation perfect, so some minor errors are allowed to occur. Errors are obviously bad in certain applications, but in the realm of graphics minor errors in the hue of a given pixel would be a small price for the kind of performance gains reported.

Also, it seems to me that as graphics continue to progress these errors will become less and less of a concern. With more realistic graphics, the natural ability of your mind to fill in the blanks will take over, and easily overlook a few stray pixels.

Patron saint of this blog Ray Kurzweil, along with NASA, Google and a bunch of other luminaries, announced today at TED the creation of the Singularity University. The curriculum will explore the consequences of exponential growth in nano, bio, and computational fields. You can read more about it at their website Singularity University

Even in my short time as a Singularitarian (~5 years) it has been interesting to see the concepts start to gain traction. The founding of this University is an important step towards entering the general public’s consciousness. It’s also just fun to feel a little bit less crazy now that so many big institutions have gotten behind this idea.

or something like that…

Yes folks, we have another claim of world beating LED lightbulb tech. As a bonus, these scientists say that their bulb is ready to be manufactured and will cost $2.85 to produce. This is good to know, but it would be even better if they dropped a hint as to what their bulb will sell for. Currently, a 60 watt equivalent LED bulb will cost you $50, so they could price this bulb at $30 and make money if they wanted to, but at that price it wouldn’t make much of a difference in our energy use (even though it is still a smart buy versus an incandescent). Priced at $10 or less, it really could produce a HUGE drop in our electricity use.

Unfortunately, that someone is an autonomous robot.

Wired has an article here about how online companies are now using robots to fetch products in their warehouses. This brings a reduction in labor costs, but also allows the companies to save money by turning the lights out and lowering the heat.

It is odd to me that in all of the discussions about the current economic crisis, nobody ever seems to bring up the point that there is simply less for us to do. Of course, if the credit markets are flowing and everybody is happy, we can all agree to work on making each other plasma TVs. But it is easy to have a crisis of confidence when 75%+ of the labor in our economy goes towards things we don’t really need.

This is just kind of neat. A recent article in New Scientist (original here) explains how we tell the difference between someone singing and talking. Turns out, the distinction is almost totally arbitrary.

Check out this illustrative audio clip, where the repetition of a spoken phrase begins to sound like singing: http://philomel.com/mp3/phantom_words/Track_22.mp3

…more of this stuff:

Restoring sight through artificial retinas

Money quote:

Another patient, Terry, spotted the shadow of his 18 year-old son as he passed by on a sidewalk. “It was the first time I’d seen anything of him since he was 5 years old,”

If you’re interested, there was a cool video about the first prototype in this PBS show (Btw, I’m not much for crying, but if your eyes don’t well up a little when the little girl in this episode gets to hear again, then you might have already passed the Turing Test)

Really, you can’t get much deeper into the “science as magic” category than giving sight to the blind.

A recent Wired article included this line:

In a sense, you could argue that even after 100 years of moving pictures, we still don’t know what video is for.

The rest of the article is here

It’s a pretty interesting reminder of how technology influences culture.

phosphorescent powder tell *him* what to do!

Wired has cool picture essay here detailing a new procedure that can almost perfectly transpose an actor’s face onto a digital model (most recently used in The Curious Case of Benjamin Button).

This is the man who reportedly will be in charge of America’s energy policy starting January 21st:

Just amazing