uncleclover
New member
- Jul 5, 2008
- 10
- 0
- 1
"Prove you're human" fill-in boxes - new ideas needed as a.i. evolves. Would this
one work? As programs grow increasingly sophisticated and capable of dealing with imagery and words in a manner much more closely resembling organic humanity's manipulation of such things, those little "prove you're human" anti-bot boxes on web sites are needing to bone up on their game. I have an idea. For the same reason that you are probably able to read this paragraph (unless English is not you're primary written language):
"Who eevr siad taht coecrect spllelnig was ncesasry to cnoevy an ieda. In fcat spllelgin is cnosatnlty eovlvnig and mrohpnig and iended eevn the mneanigs of wrods cnaghe oevr tmie. Taht is how we cmoe by so mnay lnagauegs in teh frsit pacle. Lfie ahobers a vcuuam and canhge is cntsnaote. Remebemr, olny smthonig taht is daed can be peesrrved. "
This little trait of the human mind is something computers are probably going to be a good bit far-off from being able to do any time soon (which given the pace of modern technology gives us at LEAST until next week). Instead of two random words, have them be two words that are related to one another in-context, but with the middle letters jumbled as in the above example so that the user has to figure them out. For instance:
pnaeut btuetr
Lots of native english-reading folks should be able to tell that's "peanut butter".
Trust me, by the time auto-bots become capable of snookering their way past _that_ sort of test, spam-bot activity on the Internet is likely to be the least of our concerns, because when software is capable of understanding language to the degree where it can parse even mega-jumbled text like the above, it will have attained true sentience. Thus, it will either be transforming our civilization into a utopia, or it will be exterminating us pesky little hairless chimps. ;-)
one work? As programs grow increasingly sophisticated and capable of dealing with imagery and words in a manner much more closely resembling organic humanity's manipulation of such things, those little "prove you're human" anti-bot boxes on web sites are needing to bone up on their game. I have an idea. For the same reason that you are probably able to read this paragraph (unless English is not you're primary written language):
"Who eevr siad taht coecrect spllelnig was ncesasry to cnoevy an ieda. In fcat spllelgin is cnosatnlty eovlvnig and mrohpnig and iended eevn the mneanigs of wrods cnaghe oevr tmie. Taht is how we cmoe by so mnay lnagauegs in teh frsit pacle. Lfie ahobers a vcuuam and canhge is cntsnaote. Remebemr, olny smthonig taht is daed can be peesrrved. "
This little trait of the human mind is something computers are probably going to be a good bit far-off from being able to do any time soon (which given the pace of modern technology gives us at LEAST until next week). Instead of two random words, have them be two words that are related to one another in-context, but with the middle letters jumbled as in the above example so that the user has to figure them out. For instance:
pnaeut btuetr
Lots of native english-reading folks should be able to tell that's "peanut butter".
Trust me, by the time auto-bots become capable of snookering their way past _that_ sort of test, spam-bot activity on the Internet is likely to be the least of our concerns, because when software is capable of understanding language to the degree where it can parse even mega-jumbled text like the above, it will have attained true sentience. Thus, it will either be transforming our civilization into a utopia, or it will be exterminating us pesky little hairless chimps. ;-)