Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×

Comment Re:A pretty low requirement (Score 1) 432

The bar is "thinks like a human." It's pretty clear Watson isn't intelligent in the normal sense of the word. He couldn't even carry on an interesting conversation with you, unless your entire conversation is an attempt to search the internet.

Also, who ever said, "If a computer can beat top players at Jeopardy, it's intelligent?" Who ever said, "If a computer can play chess better than a human, it's intelligent?" The Turing test has been around for a long time.

Comment Re:Outdated test (Score 1) 432

In almost all Turing tests where the computer 'passed', they've had a setup with a computer and a person. The tester chatted with both of them, and couldn't figure out which one was which.

Then when they release the actual conversations, you see the computer actually wasn't too smart, but the other person was pretending to talk like a computer. What these tests actually show is that a human can convincingly pretend to chat like a computer.

Comment Hasn't this happened a bunch of times? (Score 5, Interesting) 432

Just googling a few seconds brought me to:

This article about cleverbot., which also eeked out enough votes to 'pass' a turing test.

It's all sounds just like Eliza, just put into a character with enough human limitations that you'd expect it not to string together phrases well, or keep to one topic more than a sentence.

I'd interpret it basically as an automated DJ sound board with generic text instead of movie quotes - you can certainly string a lot of folks along with even really bad ones, but that speaks more to pareidolia than anything else.

I'd classify this stage of AI closer to "parlour trick" than "might as well be human" that a lot of people think of when they hear Turing test - but that's also part of the test, to see what we consider to be human.

Ryan Fenton

Comment Re:How much have the seas risen? (Score 5, Insightful) 182

Well, it's a bit like stairs. It's really important to make sure each riser is exactly the same, because people going up and down those stairs adapt with remarkable precision to the height of the first few steps they climb. If you took a slow motion picture, you'd see their foot gliding onto each step with a scant millimeter or so to spare. A 2mm difference in all the stairs nobody will notice; a 2mm difference in one stair will trip people up, even though you can't even *see* it.

People build around flood levels the same way. They build right up to what the historical floodline is for the frequency they can tolerate. If they can tolerate one flood every ten years, they'll build right up to to the ten year floodline. But if the sea levels rise 15cm/5.5 inches, as they have since 1945 or so, that spot might be flooded every year. You can easily imagine a gravesite that was stable in its balance between sand deposition and erosion for many years "suddenly" getting washed away, although in truth the line between stable and unstable has been continually creeping up over the decades.

Understand this is not a simple situation; 5 inches of sea level rise doesn't mean suddenly lots of homes are under water everywhere around the world. But it can mean lots of homes are getting flooded in some parts of the world. It depends on local conditions and building practices. Here in Boston, for example, we have two meter tides, and massive variation between spring and neap tides, and with the direction of wind and air pressure, and we've historically built accordingly. 5 inches of sea level rise over half a century has made no noticeable difference *here*. Other places that have very low tidal amplitudes and don't experience large storms with persistent low pressure (e.g., Venice) might find a lot of stuff getting flooded after a 5 inch sea level rise.

Comment I skimmed the front page too fast. (Score 1) 65

I skimmed the front page, and misread the title to this story as "Updating the Integrated Space Pen". Intrigued at what those ambitious scamps at the Fisher Space Pen company might be up to, I skimmed the summary for links and misread the address of the linked website as "thefacepalm.com". I still have no idea what the story is actually about, but I thought I'd chip in my contribution anyway.

All in all, the start of a perfect Slashdot Sunday for me...

Comment Re:Greater per car occupancy? (Score 1) 260

So what you are saying is that Uber is not even a ride "sharing" platform so much as an enabler for unlicensed car service business? I did not know that.

I have to admit that my opinion on Uber was, so far, essentially neutral. However, if what you are saying is true - I would be inclined to reconsider and think of them as a net-negative. If they are a taxi cab - they should register and operate as one, any instant online hailing and optimal vehicle routing sauce notwithstanding.

I will vote accordingly if/when this comes up in my locality.

Uber is basically a pre-booked taxi service. They satisfy the goals of most regulation around taxi services - safety and honesty. They probably do a better job of it than taxi regulations do. The last time I checked the state does not have records of who drove who where and when BEFORE the ride takes place with a conventional taxi.

What is the point of having taxi laws in the first place? Complaining about Uber seems like complaining about automated cars eliminating speeding violations and thus ticket revenue.

For the most part the only thing taxi regulations have done is create cozy business environments for companies that provide a service that everybody avoids like the plague. I know I'd probably only hire a taxi if my life depended on it. A service like Uber has the potential to change that.

Comment Re:Third-world Jitney service (Score 1) 260

In California, for example, drivers-for-hire have to be specifically licensed, and carry $1M liability insurance. Uber provides a $100K "umbrella" for the benefit of passengers, "just in case" the driver isn't insured as required by the company. (But the required insurance level is far less than that required by the state.)

The problem is that the system is built on ancient premises and thus is incredibly inefficient, and propped up by people with vested interests.

The premise cab regulations are built on is that there is no way to know who is picking up who, when, and where. That means that if jack the ripper gets a cab license you have no way to figure out why so many people are disappearing. It also means that there are huge insurance requirements in the hope that insurance companies do something to control who gets issued insurance (outsourcing of quality control, basically).

And yet, we still end up with cabs that are of incredibly poor quality and with fairly marginal drivers in many places. That is because once so much money got made off of regulations like medallion sales there was a ton of regulatory capture.

Today that isn't the world we live in. Today it is possible to book every trip in advance, with credit card companies acting as trusted brokers to protect both consumers and suppliers, and reputation systems for everybody involved. An operator shouldn't need a million dollars in insurance, unless they're driving something the size of a bus. Just how much injury can a car possibly cause to its passengers? You don't need to have cars just driving around so that they're visible so that they can be hailed - you can summon them on demand. The pre-booking means there is a record of who is going where, and since everybody is carrying a cell phone chances are the NSA if not the local police know where everybody is at all times anyway (not that they'd actually use that data to go after such a minor public nuisance as a serial killer).

So, the regulation really needs an overhaul. Just require pre-booking of all trips. Of course, that will be opposed by anybody with a medallion, so good luck with that.

Comment Re:If only Bill Waterson inspired other cartoonist (Score 1) 119

I am not going to complain when someone wants to make a reasonable honest living. Some people like to work, and some people have talent but don't like to do the day to day grind. It is a unique type of job to have to produce a few hundred different creative products a year. Berke Brethed is another one who had a lot of talent but did not like having to fit everything into a commercial format. So he tried to break the format by doing an awesome Sunday only strip, but that did not last long. But when you have the lure of money and people who do not have to work harder than they want, or where the work can be done in committee, the carton is not going to end. The Simpon's for instance could have been cancelled a few years ago, but the actors realized they could be replaced, and I guess having work at half the rate of sitcom actors was better than having no work at all. The cost of actors is what really killed Seinfeld and Friends. But Watterson could have subcontracted out the comic, and he did not, and for that he gets a lot of credit. Of course not every comic is controlled by the writer/artists.

Comment Re:Influence? (Score 4, Insightful) 231

It's a good point, and there is some question about the idea of using pagerank on Wikipedia as a method for measuring influential people. For one thing, it has a bias towards most recent events. They used two different algorithms for ranking influence in the English version of Wikipedia. The first version ended up with this list: "Napoleon, Barack Obama, Carl Linnaeus, Elizabeth II and George W Bush." At least it's bipartisan.

Another problem with pagerank on Wikipedia is the bias towards popularity. "Frank Sinatra, Michael Jackson, Pope Pius XII, Elton John and Elizabeth II." Was Frank Sinatra more influential than Michael Jackson?

Going from that high quality single-language ranking, they tried to rank across languages. With their second algorithm, this is what they ended up with: "Adolf Hitler, Michael Jackson, Madonna (the singer) and Ludwig Van Beethoven." I really like Beethoven, but.....

If your algorithm only matches the pre-existing ranking by 50%, that might be an indication that your algorithm isn't getting good data. In fact, the scientists involved have some doubt about the quality of their research, saying: “Our analysis shows that most important historical gures across Wikipedia language editions are born in Western countries after the 17th century, and are male”

Slashdot Top Deals

A failure will not appear until a unit has passed final inspection.

Working...