Reply to thread

Message

<blockquote data-quote="436880927" data-source="post: 826657">If you&#039;re wondering about how you&#039;d be helping their own AIs determine between different objects in a picture since you have to identify them yourself and it would need to know the answer to tell you that you&#039;ve gotten it right, the truth is that the captcha system doesn&#039;t have one definitive answer that must be met.I believe that the captcha system is aware of which pieces are definitely not part of the answer, but has difficulty determining which parts are only definitely part of the answer... and so by answering the questions, hopefully correctly, it would theoretically become stronger at knowing which ones are definitely part of the answer, as opposed to &quot;almost&quot; near the correct pieces.As the AI gets stronger, people answering them &quot;almost correctly&quot; and not &quot;completely correctly&quot; according to old collected data can be used to tie people to potentially having eyesight difficulty as well. So not just the medical condition example in the original post. There&#039;s a lot of potential for tying people to different circumstances based on weeks, months or years of captcha code collection. The longer-term profile for a specific person (even if it is only a few captcha&#039;s every few months) is also very valuable.</blockquote>

[QUOTE="436880927, post: 826657"] If you're wondering about how you'd be helping their own AIs determine between different objects in a picture since you have to identify them yourself and it would need to know the answer to tell you that you've gotten it right, the truth is that the captcha system doesn't have one definitive answer that must be met. I believe that the captcha system is aware of which pieces are definitely not part of the answer, but has difficulty determining which parts are only definitely part of the answer... and so by answering the questions, hopefully correctly, it would theoretically become stronger at knowing which ones are definitely part of the answer, as opposed to "almost" near the correct pieces. As the AI gets stronger, people answering them "almost correctly" and not "completely correctly" according to old collected data can be used to tie people to potentially having eyesight difficulty as well. So not just the medical condition example in the original post. There's a lot of potential for tying people to different circumstances based on weeks, months or years of captcha code collection. The longer-term profile for a specific person (even if it is only a few captcha's every few months) is also very valuable. [/QUOTE]

Verification