<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/2.1.2" -->
<rss version="2.0" 
	xmlns:content="http://purl.org/rss/1.0/modules/content/">
<channel>
	<title>Comments on: On the Merits of Parrots</title>
	<link>http://www.novamente.net/blog/?p=9</link>
	<description>Artificial General Intelligence</description>
	<pubDate>Thu, 09 Sep 2010 06:03:36 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.1.2</generator>

	<item>
		<title>By: Gabe</title>
		<link>http://www.novamente.net/blog/?p=9#comment-85</link>
		<author>Gabe</author>
		<pubDate>Mon, 30 Jul 2007 18:24:51 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-85</guid>
					<description>What an excellent idea. What 3D and 2D virtual environments (besides Second Life, I assume) do you have in mind?</description>
		<content:encoded><![CDATA[<p>What an excellent idea. What 3D and 2D virtual environments (besides Second Life, I assume) do you have in mind?</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Paul</title>
		<link>http://www.novamente.net/blog/?p=9#comment-86</link>
		<author>Paul</author>
		<pubDate>Tue, 31 Jul 2007 11:59:15 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-86</guid>
					<description>Ben,

   I too am growing increasingly excited about your new business plan. (if only I had some capital!)  Thank you and Bruce for your reply on ImmInst.  As you stated, the 'wisdom of the masses' effect is very powerful.  What Luis' work proves, yet again, is that positive feedback architectures essentially equate to thicker pipes connecting both biological and silicon supercomputers.  Taking as entertaining an approach as possible will net Novamente the most efficient training, as well as increase interest in the underlying scientific and humanitarian bases of the project.  
   You'd better get to work though.  As I see it, you've really only got 5 years left to pull this off, or there are going to be some pissed of Mayans.</description>
		<content:encoded><![CDATA[<p>Ben,</p>
<p>   I too am growing increasingly excited about your new business plan. (if only I had some capital!)  Thank you and Bruce for your reply on ImmInst.  As you stated, the &#8216;wisdom of the masses&#8217; effect is very powerful.  What Luis&#8217; work proves, yet again, is that positive feedback architectures essentially equate to thicker pipes connecting both biological and silicon supercomputers.  Taking as entertaining an approach as possible will net Novamente the most efficient training, as well as increase interest in the underlying scientific and humanitarian bases of the project.<br />
   You&#8217;d better get to work though.  As I see it, you&#8217;ve really only got 5 years left to pull this off, or there are going to be some pissed of Mayans.</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Phillip Huggan</title>
		<link>http://www.novamente.net/blog/?p=9#comment-87</link>
		<author>Phillip Huggan</author>
		<pubDate>Tue, 31 Jul 2007 19:10:51 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-87</guid>
					<description>The parrot idea will work one day.  As you say it isn't too different from Google's pagerank or the living wikis.  But I think it would eventually plateau at the English language level of its most active human editors.  What would really be required (assuming this context) is a "Shakespeare parrot"; a parrot that can invent new words correctly.  And this may only be possible with humans editing a living library of the parrot's knowledge base.  Certainly the parrot wouldn't be able to talk novel chemistry without access to research infrastructures, or human chemists as knowledge base editors.
This sounds like an AI parrot, not AGI.  Bw-rack.</description>
		<content:encoded><![CDATA[<p>The parrot idea will work one day.  As you say it isn&#8217;t too different from Google&#8217;s pagerank or the living wikis.  But I think it would eventually plateau at the English language level of its most active human editors.  What would really be required (assuming this context) is a &#8220;Shakespeare parrot&#8221;; a parrot that can invent new words correctly.  And this may only be possible with humans editing a living library of the parrot&#8217;s knowledge base.  Certainly the parrot wouldn&#8217;t be able to talk novel chemistry without access to research infrastructures, or human chemists as knowledge base editors.<br />
This sounds like an AI parrot, not AGI.  Bw-rack.</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Bruce Klein&#8217;s Novamente Weblog &#187; Ben Goertzel = TV07</title>
		<link>http://www.novamente.net/blog/?p=9#comment-89</link>
		<author>Bruce Klein&#8217;s Novamente Weblog &#187; Ben Goertzel = TV07</author>
		<pubDate>Wed, 01 Aug 2007 01:20:52 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-89</guid>
					<description>[...] Ben Goertzel (Novamente CEO) spoke at Transvision 2007 on a panel w/ Marvin Minsky and Second Life&#8217;s Philip Rosedale. Photo credit goes to George Dvorsky, who has a nice roundup of the event.  More links on TV07 found at the Novamente blog.  Ben also gives his account - On the Merits of Parrots. [...]</description>
		<content:encoded><![CDATA[<p>[&#8230;] Ben Goertzel (Novamente CEO) spoke at Transvision 2007 on a panel w/ Marvin Minsky and Second Life&#8217;s Philip Rosedale. Photo credit goes to George Dvorsky, who has a nice roundup of the event.  More links on TV07 found at the Novamente blog.  Ben also gives his account - On the Merits of Parrots. [&#8230;]</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: BenGoertzel</title>
		<link>http://www.novamente.net/blog/?p=9#comment-259</link>
		<author>BenGoertzel</author>
		<pubDate>Fri, 31 Aug 2007 18:10:49 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-259</guid>
					<description>Philip: 
Whether the parrots plateau or not depends on the underlying AGI architecture, of course.  If they were built in a simplistic way using standard contemporary AI components, for sure they would plateau at the level of their best human conversation partners -- if indeed they would ever get to that level, which I doubt....  But we would back-end the parrots on the Novamente Cognition Engine which has been designed with a goal of intelligence at the human level and beyond.  At the same time as the parrots' intelligence would increase due to conversational interaction, it would also increase due to us progressively implementing and tuning more and more of the NCE.  The NCE contains explicit facilities for concept creation, hypothesis formation, and so forth....  I'm not considering conversational learning from humans as a substitute for a powerful AGI architecture ... just as a helper to the latter...</description>
		<content:encoded><![CDATA[<p>Philip:<br />
Whether the parrots plateau or not depends on the underlying AGI architecture, of course.  If they were built in a simplistic way using standard contemporary AI components, for sure they would plateau at the level of their best human conversation partners &#8212; if indeed they would ever get to that level, which I doubt&#8230;.  But we would back-end the parrots on the Novamente Cognition Engine which has been designed with a goal of intelligence at the human level and beyond.  At the same time as the parrots&#8217; intelligence would increase due to conversational interaction, it would also increase due to us progressively implementing and tuning more and more of the NCE.  The NCE contains explicit facilities for concept creation, hypothesis formation, and so forth&#8230;.  I&#8217;m not considering conversational learning from humans as a substitute for a powerful AGI architecture &#8230; just as a helper to the latter&#8230;</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Andrew</title>
		<link>http://www.novamente.net/blog/?p=9#comment-389</link>
		<author>Andrew</author>
		<pubDate>Sat, 15 Sep 2007 11:52:39 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-389</guid>
					<description>I just wanted to wish you all the best with creating AGI parrots and onwards to virtual twins, I'll personally look forward to that day and will probably go and join Second Life soon to prepare myself for the new world. 

Let me tell you why I want to be partnered with a virtual twin. I teach English to adults for whom English is a second or third language and what I want to be able to do is offer more lessons, which at this point in time requires hiring a sales person to source leads for me, and may require hiring an administrator to manage classes, payments etc.

Your virtual twin actually seems like it might be a good fit for my needs. It would be even better if the virtual twin could be created with a direct interface to real world email, telephone, faxes etc. Perhaps I'm being silly, but if ideas aren't talked about and promoted as benefits then many will simply disregard the opportunity as a waste of money.</description>
		<content:encoded><![CDATA[<p>I just wanted to wish you all the best with creating AGI parrots and onwards to virtual twins, I&#8217;ll personally look forward to that day and will probably go and join Second Life soon to prepare myself for the new world. </p>
<p>Let me tell you why I want to be partnered with a virtual twin. I teach English to adults for whom English is a second or third language and what I want to be able to do is offer more lessons, which at this point in time requires hiring a sales person to source leads for me, and may require hiring an administrator to manage classes, payments etc.</p>
<p>Your virtual twin actually seems like it might be a good fit for my needs. It would be even better if the virtual twin could be created with a direct interface to real world email, telephone, faxes etc. Perhaps I&#8217;m being silly, but if ideas aren&#8217;t talked about and promoted as benefits then many will simply disregard the opportunity as a waste of money.</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: UgoTrade &#187; Blog Archive &#187; Artificial General Intelligence in Second Life</title>
		<link>http://www.novamente.net/blog/?p=9#comment-408</link>
		<author>UgoTrade &#187; Blog Archive &#187; Artificial General Intelligence in Second Life</author>
		<pubDate>Mon, 24 Sep 2007 21:07:28 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-408</guid>
					<description>[...] the Novamente blog Ben has a very interesting post On the Merits of Parrots … or: “The Wisdom of Crowds” as a [...]</description>
		<content:encoded><![CDATA[<p>[&#8230;] the Novamente blog Ben has a very interesting post On the Merits of Parrots … or: “The Wisdom of Crowds” as a [&#8230;]</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Justin Gibbs</title>
		<link>http://www.novamente.net/blog/?p=9#comment-412</link>
		<author>Justin Gibbs</author>
		<pubDate>Tue, 02 Oct 2007 02:40:41 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-412</guid>
					<description>The parrot idea would work great but I think you would get more traction if you built it around character more then relying on people's interest and desire to correct someone's language skills. If the parrot is nothing more then a data collection point my interest in it would fade  as I realized it was pretty good at natural language comprehension. Now if you gave the parrot some character I might stick around for hours. If the parrot threw out a few smart alick lines or had a chip on its shoulder I'd keeping talking to see what else it might say - I might literally keep talking just to "get to know" the parrot.

And if you really want some data, make the parrot a female bar tender. Everyone would push her for sex, but if she could avert such talk in interesting ways you would have endless data. Even better, make her a beer girl and fund the project through advertising dollars.</description>
		<content:encoded><![CDATA[<p>The parrot idea would work great but I think you would get more traction if you built it around character more then relying on people&#8217;s interest and desire to correct someone&#8217;s language skills. If the parrot is nothing more then a data collection point my interest in it would fade  as I realized it was pretty good at natural language comprehension. Now if you gave the parrot some character I might stick around for hours. If the parrot threw out a few smart alick lines or had a chip on its shoulder I&#8217;d keeping talking to see what else it might say - I might literally keep talking just to &#8220;get to know&#8221; the parrot.</p>
<p>And if you really want some data, make the parrot a female bar tender. Everyone would push her for sex, but if she could avert such talk in interesting ways you would have endless data. Even better, make her a beer girl and fund the project through advertising dollars.</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: lordvolton</title>
		<link>http://www.novamente.net/blog/?p=9#comment-447</link>
		<author>lordvolton</author>
		<pubDate>Mon, 12 Nov 2007 07:37:43 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-447</guid>
					<description>Parallel learning is one of the more interesting consequences of AI. And it's the reason most people do not fully appreciate how quickly AIs will learn, since all of us learn sequentially. 

One way to move this forward more quickly is to offer it as a "widget" on Facebook where people can teach the parrot facts about itself. So when I visit someone's facebook page the Parrot can tell me things I'd like to know.

And if I know the person I can teach the parrot interesting things about the person, so that the parrot's identity is tied to something that people will find interesting: themselves.

Others might want the parrot to have its own identity... but my guess is that a parrot that can introduce you to the relevant facts about a person would be very popular. 

This would also be useful for dating sites such as eHarmony. Imagine how much time people could save by spending time with a parrot learning what a person likes and dislikes before investing a lot of time and money.

Food for thought.

-LV</description>
		<content:encoded><![CDATA[<p>Parallel learning is one of the more interesting consequences of AI. And it&#8217;s the reason most people do not fully appreciate how quickly AIs will learn, since all of us learn sequentially. </p>
<p>One way to move this forward more quickly is to offer it as a &#8220;widget&#8221; on Facebook where people can teach the parrot facts about itself. So when I visit someone&#8217;s facebook page the Parrot can tell me things I&#8217;d like to know.</p>
<p>And if I know the person I can teach the parrot interesting things about the person, so that the parrot&#8217;s identity is tied to something that people will find interesting: themselves.</p>
<p>Others might want the parrot to have its own identity&#8230; but my guess is that a parrot that can introduce you to the relevant facts about a person would be very popular. </p>
<p>This would also be useful for dating sites such as eHarmony. Imagine how much time people could save by spending time with a parrot learning what a person likes and dislikes before investing a lot of time and money.</p>
<p>Food for thought.</p>
<p>-LV</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: journalist</title>
		<link>http://www.novamente.net/blog/?p=9#comment-566</link>
		<author>journalist</author>
		<pubDate>Fri, 08 Feb 2008 15:04:05 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-566</guid>
					<description>Hello,

Very interesting idea, has it been done in second life yet ?
I've send you an e-mail, I'm planning to write an article about your project in the french newspaper Le Monde, so if you have a few minutes to plan a skype or e-mail exchange with me it will be great. Good luck for the parrots,
Severine F.</description>
		<content:encoded><![CDATA[<p>Hello,</p>
<p>Very interesting idea, has it been done in second life yet ?<br />
I&#8217;ve send you an e-mail, I&#8217;m planning to write an article about your project in the french newspaper Le Monde, so if you have a few minutes to plan a skype or e-mail exchange with me it will be great. Good luck for the parrots,<br />
Severine F.</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Life Insurance blog</title>
		<link>http://www.novamente.net/blog/?p=9#comment-575</link>
		<author>Life Insurance blog</author>
		<pubDate>Wed, 05 Mar 2008 22:13:20 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-575</guid>
					<description>&lt;strong&gt;the insurance companies don't want you to know&lt;/strong&gt;

Information on the life insurance industry</description>
		<content:encoded><![CDATA[<p><strong>the insurance companies don&#8217;t want you to know</strong></p>
<p>Information on the life insurance industry</p>
]]></content:encoded>
				</item>
	<item>
		<title>By: Direct interaction with AI a defining feature of virtual worlds &#124; Justin Gibbs</title>
		<link>http://www.novamente.net/blog/?p=9#comment-643</link>
		<author>Direct interaction with AI a defining feature of virtual worlds &#124; Justin Gibbs</author>
		<pubDate>Tue, 19 Aug 2008 14:08:07 +0000</pubDate>
		<guid>http://www.novamente.net/blog/?p=9#comment-643</guid>
					<description>[...] worlds to gather feedback and accelerate AI development. It was over a year ago that he posted, On the Merits of Parrots, about distributing AI parrots around virtual worlds so users could interact with them and [...]</description>
		<content:encoded><![CDATA[<p>[&#8230;] worlds to gather feedback and accelerate AI development. It was over a year ago that he posted, On the Merits of Parrots, about distributing AI parrots around virtual worlds so users could interact with them and [&#8230;]</p>
]]></content:encoded>
				</item>
</channel>
</rss>
