Slashdot Mirror
Games With A Purpose Help With Tasks That Tax Computers
Falkkin writes "Luis von Ahn and his team at Carnegie Mellon University have launched GWAP, a new web site for 'Games With A Purpose.' By playing these online games, humans help provide data for problems that are hard for computers to solve, such as computer vision and sound classification. Slashdot has previously covered other human computation projects by Dr. von Ahn, including the ESP Game and reCAPTCHA. The new web site contains a re-vamping of the ESP Game as well as four completely new games." (Falkkin also points to an AP story on GWAP and to coverage at the BBC.)
Entertainment is a purpose.
I was going to give it a try, but it requires registration and I really just can't be bothered to register.
1984 was not supposed to be an instruction manual.
"Everyone benefits! Now a search engine will have a better idea of what's in those images."
What search engine? Is the information I provide to them free for anyone to use, or is it just for them?
"You play the games. Computers get smarter. Everyone benefits!"
Yeah, that's EXACTLY the kind of line I'd expect Skynet to use.
"Just a fox, a whisper."
Isn't this similar to using porn to solve CAPTCHAs? Or how about the Google image Labeler? And for a literary example, this is one part of the plot in Ender's Game, but not as obvious, and a more nefarious.
Using a large amount of real intelligence can make some problems easier, if a human can do it much easier, and some amount of noise is acceptable in the output.
If I have nothing to hide, don't search me
After playing with the site for a while, I especially like Squigl... basically you and a partner draw an outline around an object in an image. If you play, make sure you check the "auto-submit when done" button, it helps save precious time :)
Isn't this initiative similar to that launched by Chris McKintrey and Pushpinder Singh, both of whom created databases where questions used to aid in trying to give robots personality?
Didn't both of these projects fail for the same reason?
I'm sure that Slashdot actually covered this story, but for those interested, the link is here.
You missed the point. You can't just magically automate something like object recognition. You can, however, train filters on computers based on how humans identify objects in images.
eclecti.cc
You can't take the sky from me...
I'll be impressed when computers are able to tag images without using anything learned from correctly tagged data to do so.
I'll be impressed when humans are able to tag images without using anything learned from correctly tagged data to do so.
The author has an excellent Google Tech Talk where he describes exactly how these games help computers get smarter. Basically, language is a shared set of common knowledge that a single researcher cannot accurately recreate without interviewing everyone. The games are tools to do exactly that. They generate datasets for analysis, and for further game playing. For example, you might find yourself describing the word "preserve" and start with marking it the opposite of destroy, while the partner guesses "strawberry" and "raspberry".
He's got lots of neat results from that system in the talk. Go watch it.
I Browse at +4 Flamebait
Open Source Sysadmin
...are they fun?