Software has a long way to go before it is sentient. But that doesn't stop Google and other companies from developing software that intentionally adds a human component to make up for what their software lacks.
Take for instance how face recognition software works; scientists have been developing face recognition since the 60s, and it is still not perfect. Security personnel and government agencies have been utilizing the technology to help with volume, but just a few years ago there were quite a few dismal failures.
Can you trust it even now, 4-5 years later?
or how about how Google's image labeler leverages human input? Google knows the limitations of image recognition by software and so uses human eyes to fill in the gaps. By making it a game, Google has harnessed free labor in providing useful labels to random bits and bytes. And I would trust this a little more, probably because it doesn't involve a system that plays with people's lives, like some dead, robot-like process like immigration. Plus, a few people wanting to sabotage results can't hurt the overall accuracy of the system with thousands of users ensuring that accuracy.
We still have a long way to go in the manner of artificial intelligence.
Then again, look how far things have gone since the 80s :)
Subscribe to:
Post Comments (Atom)
3 comments:
Loved the article on facial recognition. I guess 61% success rate is not good enough, especially if ever used to prosecute.
Apparently software can create a unique facial image using 80 facial nodal points. I wonder if these are like chakra points that can relieve my sinus pressure?
I forgot to comment on the SAT word sentient. When computers become conscious......
MATRIX 10 - UNGRIDDED
Fascinating topic, and the U.K. is way ahead of us on the use of cameras and FRSoftware. The British caught the Tube bombers almost right away from all of the different camera angles taken from all the cameras set up in the Tube stations. Definitely a technology that can fight the war on terror!
Post a Comment