<div class="gmail_quote">On Tue, Feb 15, 2011 at 4:41 PM, BillK <span dir="ltr"><<a href="mailto:pharos@gmail.com">pharos@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div class="im">On Tue, Feb 15, 2011 at 7:56 PM, David Lubkin wrote:<br>
> You don't seem to have read what I wrote. The only question I raised about<br>
> Watson's current capabilities was whether it had a module to analyze its<br>
> failures and hone itself. *That* has been possible in software for several<br>
> decades.<br><br></div></blockquote><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;"><div class="im"></div>
No, because Watson doesn't have time to do any learning or<br>
optimisation while the game is actually in progress. Watson doesn't<br>
take any notice of opponents answers. That's why it gave the same<br>
wrong answer as an opponent had already given.<br></blockquote><div><br></div><div>According to the NOVA show, Watson does learn from opponents *correct* answers. They showed an example where the answers were supposed to be month names. Watson guessed wrong on the first question, but after a couple humans answered with month names, it correctly answered one, too. I guess they just don't have time to get feedback to Watson on wrong answers during a single question. </div>
<div><br></div><div>-Dave</div><div><br></div></div>