Yahoo captcha cracked
How long will it be before any Turing test is vulnerable to normal computers let alone an AI or semi autonomous knowbot?
5 votes  by tyrhaynes    4 comments   
(Sign in to add comments)


Comments

tortoise_74    2 votes   Monday, February 11, 2008 at 2:41 AM
There is a system for identifying yourself as the owner of a website. It is called indigo or violet or some such thing. I forget the URL. This is not of general use as it requires you to own a website but in principle its extensible. Of course, it doesn't prove your human just that the bot came via your website.
[Reply]
wal    1 votes   Monday, February 11, 2008 at 1:31 PM
You're probably referring to Open ID? But that won't help much, as you said. The problem is that bots outnumber humans by a huge margin, and slowly it's getting harder and harder to detect them as they're getting smarter. I know friends doing AI research to create software that can reply to forum comments intelligently. It's not hard to guess that once such technologies become mainstream, they'll be used by spammers to fool people and make recommendations about products with a persuasive language.

[Reply]
tortoise_74    1 votes   Monday, February 11, 2008 at 2:25 AM
[Deleted by the author.]
wal    1 votes   Saturday, February 09, 2008 at 5:52 AM
Captachs are a temporary solution and their time has passed. And I doubt any other approach would last long either. We're already pushing the limit on people's pattern recognition abilities. This as an opportunity for someone to introduce an alternative system that authenticates users based on real-world documents (dirver license, credit card, ...etc) and issues credintials that can be used on other sites. Basically, a way to be able to login to sites without giving them all my personal information, yet be able to prove that I'm human.
[Reply]
Back to:
On Singularity


Join the discussion. Sign up and add your comments.