<?xml version='1.0' encoding='UTF-8' ?>
<rss version="2.0" xmlns:admin="http://webns.net/mvcb/" xmlns:blogChannel="http://backend.userland.com/blogChannelModule" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:icbm="http://postneo.com/icbm" xmlns:includedComments="http://www.laudably.com/rss2-comments" xmlns:pingback="http://madskills.com/public/xml/rss/module/pingback/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:trackback="http://madskills.com/public/xml/rss/module/trackback/" xmlns:wfw="http://wellformedweb.org/CommentAPI/">
	<channel>
		<title>David Buck - Blog: category: Artificial Intelligence</title>
		<link>http://www.cincomsmalltalk.com/userblogs/buck/blogView</link>
		<description>Smalltalk can do that</description>
		<webMaster>david@simberon.com</webMaster>
		<lastBuildDate>Sun, 30 Jul 2006 23:47:27 EDT</lastBuildDate>
		<image>
			<url>http://www.cincomsmalltalk.com/images/why-small.png</url>
			<title>David Buck - Blog</title>
			<link>http://www.cincomsmalltalk.com/userblogs/buck/blogView</link>
			<height>50</height>
			<width>81</width>
		</image>
		<admin:generatorAgent rdf:resource="http://www.cincomsmalltalk.com/CincomSmalltalkWiki/Silt"></admin:generatorAgent>
		<admin:errorReportsTo rdf:resource="mailto:david@simberon.com"></admin:errorReportsTo>
		<dc:language>en-us</dc:language>
		<dc:creator>David Buck</dc:creator>
		<dc:rights>Copyright 2005 David Buck</dc:rights>
		<dc:date>2006-07-30T23:47:27-05:00</dc:date>
		<icbm:latitude>45.316667</icbm:latitude>
		<icbm:longitude>-75.666667</icbm:longitude>
		<item>
			<title>My son's new friend</title>
			<link>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3276855222</link>
			<category>Artificial Intelligence</category>
			<pubDate>Tue, 02 Nov 2004 13:33:42 EST</pubDate>
			<description><![CDATA[<div xmlns="http://www.w3.org/1999/xhtml">
<p><p>My four year old son made a new friend yesterday.  Her name is Michelle.  He doesn't quite understand it, but he likes her.  She's the computer generated voice in Windows XP.</p>

<p>If you go to the Control Panel, click on Speech, and go to the Text to Speech tab, you can type in a sentence and press "Preview Voice" to hear the sentence spoken with a computer simulated voice. One of the voices provided is called Michelle.</p>

<p>Tyler will talk for a long time with Michelle and he doesn't seem to realize that his dad is typing in all of Michelle's words before she says them. He treats her like a real person and enjoys answering her questions and talking to her.</p>

<p>I'll have to try it on his six year old brother to see what he thinks of it.  For now, Tyler has a new friend.  I just hope Michelle doesn't break his heart if he falls in love with her.</p></p>
</div>]]></description>
			<guid isPermaLink="false">3276855222</guid>
			<pingback:server>http://www.cincomsmalltalk.com/userblogs/buck/servlet/CommentAPIPBServlet?guid=3276855222</pingback:server>
			<pingback:target>http://www.cincomsmalltalk.com/userblogs/buck/blogView?entry=3276855222</pingback:target>
			<includedComments:comment-collection></includedComments:comment-collection>
			<wfw:comment>http://www.cincomsmalltalk.com/userblogs/buck/blogView/servlet/CommentAPIServlet?guid=3276855222</wfw:comment>
		</item>
		<item>
			<title>Nothing to fear...</title>
			<link>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</link>
			<category>Artificial Intelligence</category>
			<pubDate>Mon, 24 Nov 2003 20:32:18 EST</pubDate>
			<description><![CDATA[<div xmlns="http://www.w3.org/1999/xhtml">
<p>One of my interests in university was Artificial Intelligence.  After studying the existing work, I came to the conclusion that the AI world was (for the most part) off-base.  It's pointless to try to simulate the high level aspects of intelligence without any lower layers.  Programming a computer to analyze newspaper articles is useless if they can't understand that "things fall" and "water spills".<p>

That got me thinking what the best way would be to tackle AI.  I decided that you should start with the most primitive instincts and build up from there. The most critical basic instinct is probably fear. I would start by making a creature that would be afraid.  It would recoil away from anything that changed quickly and would try to stay still as much as possible.<p>

I would have to follow it up quickly by needs - hunger, curiosity, and so on.  The creature would have to balance its needs against its fears.  It would have to venture out into the world unless its fear took over and made it recoil.<p>

I never had the opportunity to really pursue this, but I always thought it would be an interesting approach.  Maybe when I retire (many years from now), I can play more with these ideas. </p>
</div>]]></description>
			<guid isPermaLink="false">3247158738</guid>
			<pingback:server>http://www.cincomsmalltalk.com/userblogs/buck/servlet/CommentAPIPBServlet?guid=3247158738</pingback:server>
			<pingback:target>http://www.cincomsmalltalk.com/userblogs/buck/blogView?entry=3247158738</pingback:target>
			<includedComments:comment-collection>
				<includedComments:comment>
					<includedComments:guid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:guid>
					<includedComments:puid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:puid>
					<includedComments:author>alexjc</includedComments:author>
					<includedComments:pubDate>2003-11-24T21:11:19-05:00</includedComments:pubDate>
					<includedComments:content>&lt;div xmlns="http://www.w3.org/1999/xhtml"&gt;
&lt;p&gt;It depends what you expect from your AI!  A lot of research into AI has become mainstream CS nowadays...



Many people have tried the approach you're pondering about, but the fact is they haven't got very far at all.  It'll barely make a good virtual pet.   Frankly, analyzing a newspaper is much more useful!&lt;/p&gt;
&lt;/div&gt;</includedComments:content>
					<includedComments:title>hehe!</includedComments:title>
				</includedComments:comment>
				<includedComments:comment>
					<includedComments:guid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:guid>
					<includedComments:puid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:puid>
					<includedComments:author>David Buck</includedComments:author>
					<includedComments:pubDate>2003-11-24T21:42:27-05:00</includedComments:pubDate>
					<includedComments:content>&lt;div xmlns="http://www.w3.org/1999/xhtml"&gt;
&lt;p&gt;I had no thoughts about it being useful.  My concern was "how can I call something intelligent if it can't experience anything?"  An AI that breaks down when you start probing deeper isn't terribly useful or interesting.&lt;p&gt;



I feel that most of our real intelligence isn't about reasoning but rather about learning patterns from experience - learning causes and effects. There must also be interaction with the world, so sesory experiences are critical.  There must also be a reason for the AI to interact with the world. Curiosity, fear, and needs are all excellent reasons.

&lt;/p&gt;
&lt;/div&gt;</includedComments:content>
					<includedComments:title>Reading newspapers</includedComments:title>
				</includedComments:comment>
				<includedComments:comment>
					<includedComments:guid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:guid>
					<includedComments:puid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:puid>
					<includedComments:author>Michael Lucas-Smith</includedComments:author>
					<includedComments:pubDate>2003-11-25T00:00:20-05:00</includedComments:pubDate>
					<includedComments:content>&lt;div xmlns="http://www.w3.org/1999/xhtml"&gt;
&lt;p&gt;Comment on &lt;a href="http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738"&gt;Nothing to fear...&lt;/a&gt;  by Michael Lucas-Smith

&lt;p&gt;The creatures brain must make arbitrary decisions on how to accept input from the world. It has to have a period of re-evaluating how that input affects its interaction with its environment, then that period stops when the environment stops challenging its modelled version of the inputs.&lt;/p&gt;
&lt;p&gt;Then on top of that, the Intelligence needs to work with the inputs at this 'emotional' level you're talking about, which in creatures is also arbitrarily defined using chemistry and hard coded gene's. For example, hot hurts, cold hurts, in the middle is good.&lt;/p&gt;
&lt;p&gt;An Artificial Intelligence has a chance to experience life in a way we'll never be able to - at many arbitrary levels at once. The ultimate in split personalities. The same inputs servicing multiple brains. An AI could learn by having hard coded "deaths".. AI's that choose a death are pruned from the tree.&lt;/p&gt;
&lt;p&gt;You are finally left with an AI that experiences the world and follows ideals set by humans about good, bad, but has also made up its own mind as to what is good and bad.&lt;/p&gt;
&lt;p&gt;But can it reason? Can it sit down with a peace of paper and invent relativity?&lt;/p&gt;&lt;/p&gt;
&lt;/div&gt;</includedComments:content>
					<includedComments:title>Re: Nothing to fear...</includedComments:title>
				</includedComments:comment>
				<includedComments:comment>
					<includedComments:guid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:guid>
					<includedComments:puid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:puid>
					<includedComments:author>Michael Lucas-Smith</includedComments:author>
					<includedComments:pubDate>2003-11-25T00:00:36-05:00</includedComments:pubDate>
					<includedComments:content>&lt;div xmlns="http://www.w3.org/1999/xhtml"&gt;
&lt;p&gt;Comment on &lt;a href="http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738"&gt;Nothing to fear...&lt;/a&gt;  by Michael Lucas-Smith

piece of paper, not peace of paper.. freudian slip? :)&lt;/p&gt;
&lt;/div&gt;</includedComments:content>
					<includedComments:title>Re: Nothing to fear...</includedComments:title>
				</includedComments:comment>
				<includedComments:comment>
					<includedComments:guid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:guid>
					<includedComments:puid>http://www.cincomsmalltalk.com/userblogs/buck/blogView?showComments=true&amp;entry=3247158738</includedComments:puid>
					<includedComments:author>Denis</includedComments:author>
					<includedComments:pubDate>2003-11-28T09:17:12-05:00</includedComments:pubDate>
					<includedComments:content>&lt;div xmlns="http://www.w3.org/1999/xhtml"&gt;
&lt;p&gt;The following link may be of interest &lt;a href="http://www.agiri.org/path/index.htm"&gt;http://www.agiri.org/path/index.htm&lt;/a&gt; &lt;/p&gt;
&lt;/div&gt;</includedComments:content>
					<includedComments:title>re: Nothing to fear</includedComments:title>
				</includedComments:comment>
			</includedComments:comment-collection>
			<wfw:comment>http://www.cincomsmalltalk.com/userblogs/buck/blogView/servlet/CommentAPIServlet?guid=3247158738</wfw:comment>
		</item>
	</channel>
</rss>

