Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
DEAL: For $25 - Add A Second Phone Number To Your Smartphone for life! Use promo code SLASHDOT25. Also, Slashdot's Facebook page has a chat bot now. Message it for stories and more. Check out the new SourceForge HTML5 Internet speed test! ×

Submission + - The Dark Secret at the Heart of AI (technologyreview.com)

schwit1 writes: No one really knows how the most advanced algorithms do what they do. That could be a problem.

Last year, a strange self-driving car was released onto the quiet roads of Monmouth County, New Jersey. The experimental vehicle, developed by researchers at the chip maker Nvidia, didn’t look different from other autonomous cars, but it was unlike anything demonstrated by Google, Tesla, or General Motors, and it showed the rising power of artificial intelligence. The car didn’t follow a single instruction provided by an engineer or programmer. Instead, it relied entirely on an algorithm that had taught itself to drive by watching a human do it.

Getting a car to drive this way was an impressive feat. But it’s also a bit unsettling, since it isn’t completely clear how the car makes its decisions. Information from the vehicle’s sensors goes straight into a huge network of artificial neurons that process the data and then deliver the commands required to operate the steering wheel, the brakes, and other systems. The result seems to match the responses you’d expect from a human driver. But what if one day it did something unexpected—crashed into a tree, or sat at a green light? As things stand now, it might be difficult to find out why. The system is so complicated that even the engineers who designed it may struggle to isolate the reason for any single action. And you can’t ask it: there is no obvious way to design such a system so that it could always explain why it did what it did.

The mysterious mind of this vehicle points to a looming issue with artificial intelligence. The car’s underlying AI technology, known as deep learning, has proved very powerful at solving problems in recent years, and it has been widely deployed for tasks like image captioning, voice recognition, and language translation. There is now hope that the same techniques will be able to diagnose deadly diseases, make million-dollar trading decisions, and do countless other things to transform whole industries.

But this won’t happen—or shouldn’t happen—unless we find ways of making techniques like deep learning more understandable to their creators and accountable to their users. Otherwise it will be hard to predict when failures might occur—and it’s inevitable they will. That’s one reason Nvidia’s car is still experimental.

To be fair, we don’t really understand how natural intelligence works, either.

Comment Re:Probably a minor oversight. Will likely be fixe (Score 1) 236

Please do not knock Emacs.

Emacs is very popular. Popularity seems to correlate highly with the set of users who once started up Emacs, were unable to figure out how to exit from Emacs, then had no choice but to write Emacs Lisp extensions to accomplish all other necessary tasks.

I've always wondered how Emacs became my favorite operating system.

Comment Only 1% speed of light? That sucks! (Score 1) 124

I was thinking, "this is great! I can go hang out there for a few days and come back to Earth years from now".

Unfortunately, 1% only gives a time dilation of about 1.01

    t' = t/sqrt(1 -v2/c2)

Really, you have to get to well over 90% the speed of light if you want Trump's presidency to be over in a few hours.

Comment Re:we can't even be bothered to get that right.... (Score 1) 195

I'm going to go way out on a limb here and postulate that a trip AROUND the moon is going to be something more than 477,000 miles.

I think you identified the wrong problem. "Around the moon" is fine with 300,000 to 400,000. It is the "loop back" part that is the issue.

(Maybe they just stop when they get around the moon and wait for the Earth to swing by and pick them up?)

Comment Beer of the past was sweeter? (Score 2) 109

' The beer "looked like porridge and tasted sweeter and fruitier than the clear, bitter beers of today," '

Do you know how many types of beers there are today? Just go to any local microbrewery (well, maybe not in Germany- beer purity laws and all) and you will find 3-10 very different beers that are completely different from the next microbrewery.

So, someone 5000 years from now finds a beer recipe from some "ancient" brewery and concludes all our beer tastes like PBR (Pabst Blue Ribbon).

(OK, a little beer porridge might be fun to try)

Submission + - Are Gates, Musk Being "Too Aggressive" With AI Concerns? (xconomy.com)

gthuang88 writes: Bill Gates and Elon Musk are sounding the alarm “too aggressively” over artificial intelligence’s potential negative consequences for society, says MIT professor Erik Brynjolfsson. The co-author of “The Second Machine Age” argues it will take at least 30 to 50 years for robots and software to eliminate the need for human laborers. In the meantime, he says, we should be investing in education so that people are prepared for the jobs of the future, and are focused on where they still have an advantage over machines---creativity, empathy, leadership, and teamwork.

Comment Re:I'm missing something. Leap seconds .. (Score 3, Informative) 140

The article says 2 milliseconds per century, but we've already added 27 leap seconds since 1972 ..

So, what am I missing..?

Leap seconds (and leap years) are due to the number of rotations of the earth in a year not being exactly 365. If you think about it, why would they be. There is no reason the earth should rotate exactly 365 time in one trip around the sun. So, to keep December from gradually creeping into the summer, we have to fiddle a bit with the calendar.

In this case, we are just talking about the rotation speed of the earth. In a closed system, the rotation speed can be changed by moving the mass around the Earth, such as from the equator to the poles or from the the earth surface to under the surface. Changing the shape of the Earth is essentially the same as moving the mass around so would also affect speed (going from sphere to pair shape to ellipsoid).

Since we don't live in a closed system (e.g. the moon is out there), one can also change the earths rotation speed by adding (or in our case with the moon) taking away momentum. So this also affects the rotation speed.

Submission + - Remarkable New Theory Says There's No Gravity (bigthink.com)

Jeff Socia writes: Gravity is something all of us are familiar with from our first childhood experiences. You drop something — it falls. And the way physicists have described gravity has also been pretty consistent — it’s considered one of the four main forces or “interactions” of nature and how it works has been described by Albert Einstein’s general theory of relativity all the way back in 1915.

But Professor Erik Verlinde, an expert in string theory from the University of Amsterdam and the Delta Institute of Theoretical Physics, thinks that gravity is not a fundamental force of nature because it's not always there. Instead it’s “emergent” — coming into existence from changes in microscopic bits of information in the structure of spacetime.

Comment Translation for scientists and enginers (Score 1) 140

"The split second gained since the first world war may not seem much, but the time it takes for a sunbeam to travel 600km towards Earth can cost an Olympic gold medal, "

What he/she said was, "Being second by a small amount of time will cause you to not be first"

I'm pretty sure being second by any amount of time, large or small, will cause you to not be first too.

Comment Re:"new"? huh (Score 2) 78

The news is about a cube root, not a square root.

While you are right that the cube root is interesting, the Sieve of Eratosthenes is a different algorithm. Historically, you need an array of bits, with each bit set if it is the second bit after 2, the third bit after 3, the 5th bit after 5, etc. It is WAY faster than testing each number with the factors from 2 to the square root of the number.

As I understood it, rather than needing 1000 bits for the first 1000 numbers, you need cuberoot(1000) bits or 10 bits. They also say 1/5 the bits - so maybe someone can clarify that or post a better link.

(yeah, you can optimize a bit by ignoring the even numbers and so on but its still doesn't get you close to the cube root of memory).

Slashdot Top Deals

The decision doesn't have to be logical; it was unanimous.

Working...