annotate org/sloman-old.html @ 161:8f44e4061701

s.
author Robert McIntyre <rlm@mit.edu>
date Mon, 14 Mar 2016 07:21:07 -0700
parents 414a10d51d9f
children
rev   line source
rlm@109 1 <?xml version="1.0" encoding="utf-8"?>
rlm@109 2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
rlm@109 3 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
rlm@109 4 <html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
rlm@109 5 <head>
rlm@109 6 <title>Transcript of Aaron Sloman - Artificial Intelligence - Psychology - Oxford Interview</title>
rlm@109 7 <meta http-equiv="Content-Type" content="text/html;charset=utf-8"/>
rlm@109 8 <meta name="title" content="Transcript of Aaron Sloman - Artificial Intelligence - Psychology - Oxford Interview"/>
rlm@109 9 <meta name="generator" content="Org-mode"/>
rlm@109 10 <meta name="generated" content="2013-10-04 18:49:53 UTC"/>
rlm@109 11 <meta name="author" content="Dylan Holmes"/>
rlm@109 12 <meta name="description" content=""/>
rlm@109 13 <meta name="keywords" content=""/>
rlm@109 14 <style type="text/css">
rlm@109 15 <!--/*--><![CDATA[/*><!--*/
rlm@109 16 html { font-family: Times, serif; font-size: 12pt; }
rlm@109 17 .title { text-align: center; }
rlm@109 18 .todo { color: red; }
rlm@109 19 .done { color: green; }
rlm@109 20 .tag { background-color: #add8e6; font-weight:normal }
rlm@109 21 .target { }
rlm@109 22 .timestamp { color: #bebebe; }
rlm@109 23 .timestamp-kwd { color: #5f9ea0; }
rlm@109 24 .right {margin-left:auto; margin-right:0px; text-align:right;}
rlm@109 25 .left {margin-left:0px; margin-right:auto; text-align:left;}
rlm@109 26 .center {margin-left:auto; margin-right:auto; text-align:center;}
rlm@109 27 p.verse { margin-left: 3% }
rlm@109 28 pre {
rlm@109 29 border: 1pt solid #AEBDCC;
rlm@109 30 background-color: #F3F5F7;
rlm@109 31 padding: 5pt;
rlm@109 32 font-family: courier, monospace;
rlm@109 33 font-size: 90%;
rlm@109 34 overflow:auto;
rlm@109 35 }
rlm@109 36 table { border-collapse: collapse; }
rlm@109 37 td, th { vertical-align: top; }
rlm@109 38 th.right { text-align:center; }
rlm@109 39 th.left { text-align:center; }
rlm@109 40 th.center { text-align:center; }
rlm@109 41 td.right { text-align:right; }
rlm@109 42 td.left { text-align:left; }
rlm@109 43 td.center { text-align:center; }
rlm@109 44 dt { font-weight: bold; }
rlm@109 45 div.figure { padding: 0.5em; }
rlm@109 46 div.figure p { text-align: center; }
rlm@109 47 div.inlinetask {
rlm@109 48 padding:10px;
rlm@109 49 border:2px solid gray;
rlm@109 50 margin:10px;
rlm@109 51 background: #ffffcc;
rlm@109 52 }
rlm@109 53 textarea { overflow-x: auto; }
rlm@109 54 .linenr { font-size:smaller }
rlm@109 55 .code-highlighted {background-color:#ffff00;}
rlm@109 56 .org-info-js_info-navigation { border-style:none; }
rlm@109 57 #org-info-js_console-label { font-size:10px; font-weight:bold;
rlm@109 58 white-space:nowrap; }
rlm@109 59 .org-info-js_search-highlight {background-color:#ffff00; color:#000000;
rlm@109 60 font-weight:bold; }
rlm@109 61 /*]]>*/-->
rlm@109 62 </style>
rlm@109 63 <link rel="stylesheet" type="text/css" href="../css/sloman.css" />
rlm@109 64 <script type="text/javascript">
rlm@109 65 <!--/*--><![CDATA[/*><!--*/
rlm@109 66 function CodeHighlightOn(elem, id)
rlm@109 67 {
rlm@109 68 var target = document.getElementById(id);
rlm@109 69 if(null != target) {
rlm@109 70 elem.cacheClassElem = elem.className;
rlm@109 71 elem.cacheClassTarget = target.className;
rlm@109 72 target.className = "code-highlighted";
rlm@109 73 elem.className = "code-highlighted";
rlm@109 74 }
rlm@109 75 }
rlm@109 76 function CodeHighlightOff(elem, id)
rlm@109 77 {
rlm@109 78 var target = document.getElementById(id);
rlm@109 79 if(elem.cacheClassElem)
rlm@109 80 elem.className = elem.cacheClassElem;
rlm@109 81 if(elem.cacheClassTarget)
rlm@109 82 target.className = elem.cacheClassTarget;
rlm@109 83 }
rlm@109 84 /*]]>*///-->
rlm@109 85 </script>
rlm@109 86
rlm@109 87 </head>
rlm@109 88 <body>
rlm@109 89
rlm@109 90
rlm@109 91 <div id="content">
rlm@109 92 <h1 class="title">Transcript of Aaron Sloman - Artificial Intelligence - Psychology - Oxford Interview</h1>
rlm@109 93
rlm@109 94
rlm@109 95 <blockquote>
rlm@109 96
rlm@109 97
rlm@109 98
rlm@109 99
rlm@109 100
rlm@109 101
rlm@109 102
rlm@109 103
rlm@109 104
rlm@109 105
rlm@109 106
rlm@109 107
rlm@109 108
rlm@109 109
rlm@109 110
rlm@109 111
rlm@109 112 <p>
rlm@109 113 <b>Editor's note:</b> This is a working draft transcript which I made of
rlm@109 114 <a href="http://www.youtube.com/watch?feature=player_detailpage&amp;v=iuH8dC7Snno">this nice interview</a> of Aaron Sloman. Having just finished one
rlm@109 115 iteration of transcription, I still need to go in and clean up the
rlm@109 116 formatting and fix the parts that I misheard, so you can expect the
rlm@109 117 text to improve significantly in the near future.
rlm@109 118 </p>
rlm@109 119 <p>
rlm@109 120 To the extent that this is my work, you have my permission to make
rlm@109 121 copies of this transcript for your own purposes. Also, feel free to
rlm@109 122 e-mail me with comments or corrections.
rlm@109 123 </p>
rlm@109 124 <p>
rlm@109 125 You can send mail to <code>transcript@aurellem.org</code>.
rlm@109 126 </p>
rlm@109 127 <p>
rlm@109 128 Cheers,
rlm@109 129 </p>
rlm@109 130 <p>
rlm@109 131 &mdash;Dylan
rlm@109 132 </p>
rlm@109 133 </blockquote>
rlm@109 134
rlm@109 135
rlm@109 136
rlm@109 137
rlm@109 138
rlm@109 139 <div id="table-of-contents">
rlm@109 140 <h2>Table of Contents</h2>
rlm@109 141 <div id="text-table-of-contents">
rlm@109 142 <ul>
rlm@109 143 <li><a href="#sec-1">1 Introduction</a>
rlm@109 144 <ul>
rlm@109 145 <li><a href="#sec-1-1">1.1 Aaron Sloman evolves into a philosopher of AI</a></li>
rlm@109 146 <li><a href="#sec-1-2">1.2 AI is hard, in part because there are tempting non-problems.</a></li>
rlm@109 147 </ul>
rlm@109 148 </li>
rlm@109 149 <li><a href="#sec-2">2 What problems of intelligence did evolution solve?</a>
rlm@109 150 <ul>
rlm@109 151 <li><a href="#sec-2-1">2.1 Intelligence consists of solutions to many evolutionary problems; no single development (e.g. communication) was key to human-level intelligence.</a></li>
rlm@109 152 <li><a href="#sec-2-2">2.2 Speculation about how communication might have evolved from internal lanagues.</a></li>
rlm@109 153 </ul>
rlm@109 154 </li>
rlm@109 155 <li><a href="#sec-3">3 How do language and internal states relate to AI?</a>
rlm@109 156 <ul>
rlm@109 157 <li><a href="#sec-3-1">3.1 In AI, false assumptions can lead investigators astray.</a></li>
rlm@109 158 <li><a href="#sec-3-2">3.2 Example: Vision is not just about finding surfaces, but about finding affordances.</a></li>
rlm@109 159 <li><a href="#sec-3-3">3.3 Online and offline intelligence</a></li>
rlm@109 160 <li><a href="#sec-3-4">3.4 Example: Even toddlers use sophisticated geometric knowledge</a></li>
rlm@109 161 </ul>
rlm@109 162 </li>
rlm@109 163 <li><a href="#sec-4">4 Animal intelligence</a>
rlm@109 164 <ul>
rlm@109 165 <li><a href="#sec-4-1">4.1 The priority is <i>cataloguing</i> what competences have evolved, not ranking them.</a></li>
rlm@109 166 <li><a href="#sec-4-2">4.2 AI can be used to test philosophical theories</a></li>
rlm@109 167 </ul>
rlm@109 168 </li>
rlm@109 169 <li><a href="#sec-5">5 Is abstract general intelligence feasible?</a>
rlm@109 170 <ul>
rlm@109 171 <li><a href="#sec-5-1">5.1 It's misleading to compare the brain and its neurons to a computer made of transistors</a></li>
rlm@109 172 <li><a href="#sec-5-2">5.2 For example, brains may rely heavily on chemical information processing</a></li>
rlm@109 173 <li><a href="#sec-5-3">5.3 Brain algorithms may simply be optimized for certain kinds of information processing other than bit manipulations</a></li>
rlm@109 174 <li><a href="#sec-5-4">5.4 Example: find the shortest path by dangling strings</a></li>
rlm@109 175 <li><a href="#sec-5-5">5.5 In sum, we know surprisingly little about the kinds of problems that evolution solved, and the manner in which they were solved.</a></li>
rlm@109 176 </ul>
rlm@109 177 </li>
rlm@109 178 <li><a href="#sec-6">6 A singularity of cognitive catch-up</a>
rlm@109 179 <ul>
rlm@109 180 <li><a href="#sec-6-1">6.1 What if it will take a lifetime to learn enough to make something new?</a></li>
rlm@109 181 </ul>
rlm@109 182 </li>
rlm@109 183 <li><a href="#sec-7">7 Spatial reasoning: a difficult problem</a>
rlm@109 184 <ul>
rlm@109 185 <li><a href="#sec-7-1">7.1 Example: Spatial proof that the angles of any triangle add up to a half-circle</a></li>
rlm@109 186 <li><a href="#sec-7-2">7.2 Geometric results are fundamentally different than experimental results in chemistry or physics.</a></li>
rlm@109 187 </ul>
rlm@109 188 </li>
rlm@109 189 <li><a href="#sec-8">8 Is near-term artificial general intelligence likely?</a>
rlm@109 190 <ul>
rlm@109 191 <li><a href="#sec-8-1">8.1 Two interpretations: a single mechanism for all problems, or many mechanisms unified in one program.</a></li>
rlm@109 192 </ul>
rlm@109 193 </li>
rlm@109 194 <li><a href="#sec-9">9 Abstract General Intelligence impacts</a></li>
rlm@109 195 </ul>
rlm@109 196 </div>
rlm@109 197 </div>
rlm@109 198
rlm@109 199 <div id="outline-container-1" class="outline-2">
rlm@109 200 <h2 id="sec-1"><span class="section-number-2">1</span> Introduction</h2>
rlm@109 201 <div class="outline-text-2" id="text-1">
rlm@109 202
rlm@109 203
rlm@109 204
rlm@109 205 </div>
rlm@109 206
rlm@109 207 <div id="outline-container-1-1" class="outline-3">
rlm@109 208 <h3 id="sec-1-1"><span class="section-number-3">1.1</span> Aaron Sloman evolves into a philosopher of AI</h3>
rlm@109 209 <div class="outline-text-3" id="text-1-1">
rlm@109 210
rlm@109 211 <p>[0:09] My name is Aaron Sloman. My first degree many years ago in
rlm@109 212 Capetown University was in Physics and Mathematics, and I intended to
rlm@109 213 go and be a mathematician. I came to Oxford and encountered
rlm@109 214 philosophers &mdash; I had started reading philosophy and discussing
rlm@109 215 philosophy before then, and then I found that there were philosophers
rlm@109 216 who said things about mathematics that I thought were wrong, so
rlm@109 217 gradually got more and more involved in [philosophy] discussions and
rlm@109 218 switched to doing philosophy DPhil. Then I became a philosophy
rlm@109 219 lecturer and about six years later, I was introduced to artificial
rlm@109 220 intelligence when I was a lecturer at Sussex University in philosophy
rlm@109 221 and I very soon became convinced that the best way to make progress in
rlm@109 222 both areas of philosophy (including philosophy of mathematics which I
rlm@109 223 felt i hadn't dealt with adequately in my DPhil) about the philosophy
rlm@109 224 of mathematics, philosophy of mind, philsophy of language and all
rlm@109 225 those things&mdash;the best way was to try to design and test working
rlm@109 226 fragments of mind and maybe eventually put them all together but
rlm@109 227 initially just working fragments that would do various things.
rlm@109 228 </p>
rlm@109 229 <p>
rlm@109 230 [1:12] And I learned to program and ~ with various other people
rlm@109 231 including ~Margaret Boden whom you've interviewed, developed&mdash;helped
rlm@109 232 develop an undergraduate degree in AI and other things and also began
rlm@109 233 to do research in AI and so on which I thought of as doing philosophy,
rlm@109 234 primarily.
rlm@109 235 </p>
rlm@109 236 <p>
rlm@109 237 [1:29] And then I later moved to the University of Birmingham and I
rlm@109 238 was there &mdash; I came in 1991 &mdash; and I've been retired for a while but
rlm@109 239 I'm not interested in golf or gardening so I just go on doing full
rlm@109 240 time research and my department is happy to keep me on without paying
rlm@109 241 me and provide space and resources and I come, meeting bright people
rlm@109 242 at conferences and try to learn and make progress if I can.
rlm@109 243 </p>
rlm@109 244 </div>
rlm@109 245
rlm@109 246 </div>
rlm@109 247
rlm@109 248 <div id="outline-container-1-2" class="outline-3">
rlm@109 249 <h3 id="sec-1-2"><span class="section-number-3">1.2</span> AI is hard, in part because there are tempting non-problems.</h3>
rlm@109 250 <div class="outline-text-3" id="text-1-2">
rlm@109 251
rlm@109 252
rlm@109 253 <p>
rlm@109 254 One of the things I learnt and understood more and more over the many
rlm@109 255 years &mdash; forty years or so since I first encountered AI &mdash; is how
rlm@109 256 hard the problems are, and in part that's because it's very often
rlm@109 257 tempting to <i>think</i> the problem is something different from what it
rlm@109 258 actually is, and then people design solutions to the non-problems, and
rlm@109 259 I think of most of my work now as just helping to clarify what the
rlm@109 260 problems are: what is it that we're trying to explain &mdash; and maybe
rlm@109 261 this is leading into what you wanted to talk about:
rlm@109 262 </p>
rlm@109 263 <p>
rlm@109 264 I now think that one of the ways of getting a deep understanding of
rlm@109 265 that is to find out what were the problems that biological evolution
rlm@109 266 solved, because we are a product of <i>many</i> solutions to <i>many</i>
rlm@109 267 problems, and if we just try to go in and work out what the whole
rlm@109 268 system is doing, we may get it all wrong, or badly wrong.
rlm@109 269 </p>
rlm@109 270
rlm@109 271 </div>
rlm@109 272 </div>
rlm@109 273
rlm@109 274 </div>
rlm@109 275
rlm@109 276 <div id="outline-container-2" class="outline-2">
rlm@109 277 <h2 id="sec-2"><span class="section-number-2">2</span> What problems of intelligence did evolution solve?</h2>
rlm@109 278 <div class="outline-text-2" id="text-2">
rlm@109 279
rlm@109 280
rlm@109 281
rlm@109 282 </div>
rlm@109 283
rlm@109 284 <div id="outline-container-2-1" class="outline-3">
rlm@109 285 <h3 id="sec-2-1"><span class="section-number-3">2.1</span> Intelligence consists of solutions to many evolutionary problems; no single development (e.g. communication) was key to human-level intelligence.</h3>
rlm@109 286 <div class="outline-text-3" id="text-2-1">
rlm@109 287
rlm@109 288
rlm@109 289 <p>
rlm@109 290 [2:57] Well, first I would challenge that we are the dominant
rlm@109 291 species. I know it looks like that but actually if you count biomass,
rlm@109 292 if you count number of species, if you count number of individuals,
rlm@109 293 the dominant species are microbes &mdash; maybe not one of them but anyway
rlm@109 294 they're the ones who dominate in that sense, and furthermore we are
rlm@109 295 mostly &mdash; we are largely composed of microbes, without which we
rlm@109 296 wouldn't survive.
rlm@109 297 </p>
rlm@109 298
rlm@109 299 <p>
rlm@109 300 [3:27] But there are things that make humans (you could say) best at
rlm@109 301 those things, or worst at those things, but it's a combination. And I
rlm@109 302 think it was a collection of developments of which there isn't any
rlm@109 303 single one. [] there might be, some people say, human language which
rlm@109 304 changed everything. By our human language, they mean human
rlm@109 305 communication in words, but I think that was a later development from
rlm@109 306 what must have started as the use of <i>internal</i> forms of
rlm@109 307 representation &mdash; which are there in nest-building birds, in
rlm@109 308 pre-verbal children, in hunting mammals &mdash; because you can't take in
rlm@109 309 information about a complex structured environment in which things can
rlm@109 310 change and you may have to be able to work out what's possible and
rlm@109 311 what isn't possible, without having some way of representing the
rlm@109 312 components of the environment, their relationships, the kinds of
rlm@109 313 things they can and can't do, the kinds of things you might or might
rlm@109 314 not be able to do &mdash; and <i>that</i> kind of capability needs internal
rlm@109 315 languages, and I and colleagues [at Birmingham] have been referring to
rlm@109 316 them as generalized languages because some people object to
rlm@109 317 referring&hellip;to using language to refer to something that isn't used
rlm@109 318 for communication. But from that viewpoint, not only humans but many
rlm@109 319 other animals developed abilities to do things to their environment to
rlm@109 320 make them more friendly to themselves, which depended on being able to
rlm@109 321 represent possible futures, possible actions, and work out what's the
rlm@109 322 best thing to do.
rlm@109 323 </p>
rlm@109 324 <p>
rlm@109 325 [5:13] And nest-building in corvids for instance&mdash;crows, magpies,
rlm@109 326 [hawks], and so on &mdash; are way beyond what current robots can do, and
rlm@109 327 in fact I think most humans would be challenged if they had to go and
rlm@109 328 find a collection of twigs, one at a time, maybe bring them with just
rlm@109 329 one hand &mdash; or with your mouth &mdash; and assemble them into a
rlm@109 330 structure that, you know, is shaped like a nest, and is fairly rigid,
rlm@109 331 and you could trust your eggs in them when wind blows. But they're
rlm@109 332 doing it, and so &hellip; they're not our evolutionary ancestors, but
rlm@109 333 they're an indication &mdash; and that example is an indication &mdash; of
rlm@109 334 what must have evolved in order to provide control over the
rlm@109 335 environment in <i>that</i> species.
rlm@109 336 </p>
rlm@109 337 </div>
rlm@109 338
rlm@109 339 </div>
rlm@109 340
rlm@109 341 <div id="outline-container-2-2" class="outline-3">
rlm@109 342 <h3 id="sec-2-2"><span class="section-number-3">2.2</span> Speculation about how communication might have evolved from internal lanagues.</h3>
rlm@109 343 <div class="outline-text-3" id="text-2-2">
rlm@109 344
rlm@109 345 <p>[5:56] And I think hunting mammals, fruit-picking mammals, mammals
rlm@109 346 that can rearrange parts of the environment, provide shelters, needed
rlm@109 347 to have &hellip;. also needed to have ways of representing possible
rlm@109 348 futures, not just what's there in the environment. I think at a later
rlm@109 349 stage, that developed into a form of communication, or rather the
rlm@109 350 <i>internal</i> forms of representation became usable as a basis for
rlm@109 351 providing [context] to be communicated. And that happened, I think,
rlm@109 352 initially through performing actions that expressed intentions, and
rlm@109 353 probably led to situtations where an action (for instance, moving some
rlm@109 354 large object) was performed more easily, or more successfully, or more
rlm@109 355 accurately if it was done collaboratively. So someone who had worked
rlm@109 356 out what to do might start doing it, and then a conspecific might be
rlm@109 357 able to work out what the intention is, because that person has the
rlm@109 358 <i>same</i> forms of representation and can build theories about what's
rlm@109 359 going on, and might then be able to help.
rlm@109 360 </p>
rlm@109 361 <p>
rlm@109 362 [7:11] You can imagine that if that started happening more (a lot of
rlm@109 363 collaboration based on inferred intentions and plans) then sometimes
rlm@109 364 the inferences might be obscure and difficult, so the <i>actions</i> might
rlm@109 365 be enhanced to provide signals as to what the intention is, and what
rlm@109 366 the best way is to help, and so on.
rlm@109 367 </p>
rlm@109 368 <p>
rlm@109 369 [7:35] So, this is all handwaving and wild speculation, but I think
rlm@109 370 it's consistent with a large collection of facts which one can look at
rlm@109 371 &mdash; and find if one looks for them, but one won't know if [some]one
rlm@109 372 doesn't look for them &mdash; about the way children, for instance, who
rlm@109 373 can't yet talk, communicate, and the things they'll do, like going to
rlm@109 374 the mother and turning the face to point in the direction where the
rlm@109 375 child wants it to look and so on; that's an extreme version of action
rlm@109 376 indicating intention.
rlm@109 377 </p>
rlm@109 378 <p>
rlm@109 379 [8:03] Anyway. That's a very long roundabout answer to one conjecture
rlm@109 380 that the use of communicative language is what gave humans their
rlm@109 381 unique power to create and destroy and whatever, and I'm saying that
rlm@109 382 if by that you mean <i>communicative</i> language, then I'm saying there
rlm@109 383 was something before that which was <i>non</i>-communicative language, and I
rlm@109 384 suspect that noncommunicative language continues to play a deep role
rlm@109 385 in <i>all</i> human perception &mdash;in mathematical and scientific reasoning, in
rlm@109 386 problem solving &mdash; and we don't understand very much about it.
rlm@109 387 </p>
rlm@109 388 <p>
rlm@109 389 [8:48]
rlm@109 390 I'm sure there's a lot more to be said about the development of
rlm@109 391 different kinds of senses, the development of brain structures and
rlm@109 392 mechanisms is above all that, but perhaps I've droned on long enough
rlm@109 393 on that question.
rlm@109 394 </p>
rlm@109 395
rlm@109 396 </div>
rlm@109 397 </div>
rlm@109 398
rlm@109 399 </div>
rlm@109 400
rlm@109 401 <div id="outline-container-3" class="outline-2">
rlm@109 402 <h2 id="sec-3"><span class="section-number-2">3</span> How do language and internal states relate to AI?</h2>
rlm@109 403 <div class="outline-text-2" id="text-3">
rlm@109 404
rlm@109 405
rlm@109 406 <p>
rlm@109 407 [9:09] Well, I think most of the human and animal capabilities that
rlm@109 408 I've been referring to are not yet to be found in current robots or
rlm@109 409 [computing] systems, and I think there are two reasons for that: one
rlm@109 410 is that it's intrinsically very difficult; I think that in particular
rlm@109 411 it may turn out that the forms of information processing that one can
rlm@109 412 implement on digital computers as we currently know them may not be as
rlm@109 413 well suited to performing some of these tasks as other kinds of
rlm@109 414 computing about which we don't know so much &mdash; for example, I think
rlm@109 415 there may be important special features about <i>chemical</i> computers
rlm@109 416 which we might [talk about in a little bit? find out about].
rlm@109 417 </p>
rlm@109 418
rlm@109 419 </div>
rlm@109 420
rlm@109 421 <div id="outline-container-3-1" class="outline-3">
rlm@109 422 <h3 id="sec-3-1"><span class="section-number-3">3.1</span> In AI, false assumptions can lead investigators astray.</h3>
rlm@109 423 <div class="outline-text-3" id="text-3-1">
rlm@109 424
rlm@109 425 <p>[9:57] So, one of the problems then is that the tasks are hard &hellip; but
rlm@109 426 there's a deeper problem as to why AI hasn't made a great deal of
rlm@109 427 progress on these problems that I'm talking about, and that is that
rlm@109 428 most AI researchers assume things&mdash;and this is not just AI
rlm@109 429 researchers, but [also] philsophers, and psychologists, and people
rlm@109 430 studying animal behavior&mdash;make assumptions about what it is that
rlm@109 431 animals or humans do, for instance make assumptions about what vision
rlm@109 432 is for, or assumptions about what motivation is and how motivation
rlm@109 433 works, or assumptions about how learning works, and then they try ---
rlm@109 434 the AI people try &mdash; to model [or] build systems that perform those
rlm@109 435 assumed functions. So if you get the <i>functions</i> wrong, then even if
rlm@109 436 you implement some of the functions that you're trying to implement,
rlm@109 437 they won't necessarily perform the tasks that the initial objective
rlm@109 438 was to imitate, for instance the tasks that humans, and nest-building
rlm@109 439 birds, and monkeys and so on can perform.
rlm@109 440 </p>
rlm@109 441 </div>
rlm@109 442
rlm@109 443 </div>
rlm@109 444
rlm@109 445 <div id="outline-container-3-2" class="outline-3">
rlm@109 446 <h3 id="sec-3-2"><span class="section-number-3">3.2</span> Example: Vision is not just about finding surfaces, but about finding affordances.</h3>
rlm@109 447 <div class="outline-text-3" id="text-3-2">
rlm@109 448
rlm@109 449 <p>[11:09] I'll give you a simple example &mdash; well, maybe not so simple,
rlm@109 450 but &mdash; It's often assumed that the function of vision in humans (and
rlm@109 451 in other animals with good eyesight and so on) is to take in optical
rlm@109 452 information that hits the retina, and form into the (maybe changing
rlm@109 453 &mdash; or, really, in our case definitely changing) patterns of
rlm@109 454 illumination where there are sensory receptors that detect those
rlm@109 455 patterns, and then somehow from that information (plus maybe other
rlm@109 456 information gained from head movement or from comparisons between two
rlm@109 457 eyes) to work out what there was in the environment that produced
rlm@109 458 those patterns, and that is often taken to mean &ldquo;where were the
rlm@109 459 surfaces off which the light bounced before it came to me&rdquo;. So
rlm@109 460 you essentially think of the task of the visual system as being to
rlm@109 461 reverse the image formation process: so the 3D structure's there, the
rlm@109 462 lens causes the image to form in the retina, and then the brain goes
rlm@109 463 back to a model of that 3D structure there. That's a very plausible
rlm@109 464 theory about vision, and it may be that that's a <i>subset</i> of what
rlm@109 465 human vision does, but I think James Gibson pointed out that that kind
rlm@109 466 of thing is not necessarily going to be very useful for an organism,
rlm@109 467 and it's very unlikely that that's the main function of perception in
rlm@109 468 general, namely to produce some physical description of what's out
rlm@109 469 there.
rlm@109 470 </p>
rlm@109 471 <p>
rlm@109 472 [12:37] What does an animal <i>need</i>? It needs to know what it can do,
rlm@109 473 what it can't do, what the consequences of its actions will be
rlm@109 474 &hellip;. so, he introduced the word <i>affordance</i>, so from his point of
rlm@109 475 view, the function of vision, perception, are to inform the organism
rlm@109 476 of what the <i>affordances</i> are for action, where that would mean what
rlm@109 477 the animal, <i>given</i> its morphology (what it can do with its mouth, its
rlm@109 478 limbs, and so on, and the ways it can move) what it can do, what its
rlm@109 479 needs are, what the obstacles are, and how the environment supports or
rlm@109 480 obstructs those possible actions.
rlm@109 481 </p>
rlm@109 482 <p>
rlm@109 483 [13:15] And that's a very different collection of information
rlm@109 484 structures that you need from, say, &ldquo;where are all the
rlm@109 485 surfaces?&rdquo;: if you've got all the surfaces, <i>deriving</i> the
rlm@109 486 affordances would still be a major task. So, if you think of the
rlm@109 487 perceptual system as primarily (for biological organisms) being
rlm@109 488 devices that provide information about affordances and so on, then the
rlm@109 489 tasks look very different. And most of the people working, doing
rlm@109 490 research on computer vision in robots, I think haven't taken all that
rlm@109 491 on board, so they're trying to get machines to do things which, even
rlm@109 492 if they were successful, would not make the robots very intelligent
rlm@109 493 (and in fact, even the ones they're trying to do are not really easy
rlm@109 494 to do, and they don't succeed very well&mdash; although, there's progress;
rlm@109 495 I shouldn't disparage it too much.)
rlm@109 496 </p>
rlm@109 497 </div>
rlm@109 498
rlm@109 499 </div>
rlm@109 500
rlm@109 501 <div id="outline-container-3-3" class="outline-3">
rlm@109 502 <h3 id="sec-3-3"><span class="section-number-3">3.3</span> Online and offline intelligence</h3>
rlm@109 503 <div class="outline-text-3" id="text-3-3">
rlm@109 504
rlm@109 505
rlm@109 506 <p>
rlm@109 507 [14:10] It gets more complex as animals get more sophisticated. So, I
rlm@109 508 like to make a distinction between online intelligence and offline
rlm@109 509 intelligence. So, for example, if I want to pick something up &mdash; like
rlm@109 510 this leaf &lt;he plucks a leaf from the table&gt; &mdash; I was able to select
rlm@109 511 it from all the others in there, and while moving my hand towards it,
rlm@109 512 I was able to guide its trajectory, making sure it was going roughly
rlm@109 513 in the right direction &mdash; as opposed to going out there, which
rlm@109 514 wouldn't have been able to pick it up &mdash; and these two fingers ended
rlm@109 515 up with a portion of the leaf between them, so that I was able to tell
rlm@109 516 when I'm ready to do that &lt;he clamps the leaf between two fingers&gt;
rlm@109 517 and at that point, I clamped my fingers and then I could pick up the
rlm@109 518 leaf.
rlm@109 519 </p>
rlm@109 520 <p>
rlm@109 521 [14:54] Whereas, &mdash; and that's an example of online intelligence:
rlm@109 522 during the performance of an action (both from the stage where it's
rlm@109 523 initiated, and during the intermediate stages, and where it's
rlm@109 524 completed) I'm taking in information relevant to controlling all those
rlm@109 525 stages, and that relevant information keeps changing. That means I
rlm@109 526 need stores of transient information which gets discarded almost
rlm@109 527 immediately and replaced or something. That's online intelligence. And
rlm@109 528 there are many forms; that's just one example, and Gibson discussed
rlm@109 529 quite a lot of examples which I won't try to replicate now.
rlm@109 530 </p>
rlm@109 531 <p>
rlm@109 532 [15:30] But in offline intelligence, you're not necessarily actually
rlm@109 533 <i>performing</i> the actions when you're using your intelligence; you're
rlm@109 534 thinking about <i>possible</i> actions. So, for instance, I could think
rlm@109 535 about how fast or by what route I would get back to the lecture room
rlm@109 536 if I wanted to [get to the next talk] or something. And I know where
rlm@109 537 the door is, roughly speaking, and I know roughly which route I would
rlm@109 538 take, when I go out, I should go to the left or to the right, because
rlm@109 539 I've stored information about where the spaces are, where the
rlm@109 540 buildings are, where the door was that we came out &mdash; but in using
rlm@109 541 that information to think about that route, I'm not actually
rlm@109 542 performing the action. I'm not even <i>simulating</i> it in detail: the
rlm@109 543 precise details of direction and speed and when to clamp my fingers,
rlm@109 544 or when to contract my leg muscles when walking, are all irrelevant to
rlm@109 545 thinking about a good route, or thinking about the potential things
rlm@109 546 that might happen on the way. Or what would be a good place to meet
rlm@109 547 someone who I think [for an acquaintance in particular] &mdash; [barber]
rlm@109 548 or something &mdash; I don't necessarily have to work out exactly <i>where</i>
rlm@109 549 the person's going to stand, or from what angle I would recognize
rlm@109 550 them, and so on.
rlm@109 551 </p>
rlm@109 552 <p>
rlm@109 553 [16:46] So, offline intelligence &mdash; which I think became not just a
rlm@109 554 human competence; I think there are other animals that have aspects of
rlm@109 555 it: Squirrels are very impressive as you watch them. Gray squirrels at
rlm@109 556 any rate, as you watch them defeating squirrel-proof birdfeeders, seem
rlm@109 557 to have a lot of that [offline intelligence], as well as the online
rlm@109 558 intelligence when they eventually perform the action they've worked
rlm@109 559 out [] that will get them to the nuts.
rlm@109 560 </p>
rlm@109 561 <p>
rlm@109 562 [17:16] And I think that what happened during our evolution is that
rlm@109 563 mechanisms for acquiring and processing and storing and manipulating
rlm@109 564 information that is more and more remote from the performance of
rlm@109 565 actions developed. An example is taking in information about where
rlm@109 566 locations are that you might need to go to infrequently: There's a
rlm@109 567 store of a particular type of material that's good for building on
rlm@109 568 roofs of houses or something out around there in some
rlm@109 569 direction. There's a good place to get water somewhere in another
rlm@109 570 direction. There are people that you'd like to go and visit in
rlm@109 571 another place, and so on.
rlm@109 572 </p>
rlm@109 573 <p>
rlm@109 574 [17:59] So taking in information about an extended environment and
rlm@109 575 building it into a structure that you can make use of for different
rlm@109 576 purposes is another example of offline intelligence. And when we do
rlm@109 577 that, we sometimes use only our brains, but in modern times, we also
rlm@109 578 learned how to make maps on paper and walls and so on. And it's not
rlm@109 579 clear whether the stuff inside our heads has the same structures as
rlm@109 580 the maps we make on paper: the maps on paper have a different
rlm@109 581 function; they may be used to communicate with others, or meant for
rlm@109 582 <i>looking</i> at, whereas the stuff in your head you don't <i>look</i> at; you
rlm@109 583 use it in some other way.
rlm@109 584 </p>
rlm@109 585 <p>
rlm@109 586 [18:46] So, what I'm getting at is that there's a great deal of human
rlm@109 587 intelligence (and animal intelligence) which is involved in what's
rlm@109 588 possible in the future, what exists in distant places, what might have
rlm@109 589 happened in the past (sometimes you need to know why something is as
rlm@109 590 it is, because that might be relevant to what you should or shouldn't
rlm@109 591 do in the future, and so on), and I think there was something about
rlm@109 592 human evolution that extended that offline intelligence way beyond
rlm@109 593 that of animals. And I don't think it was <i>just</i> human language, (but
rlm@109 594 human language had something to do with it) but I think there was
rlm@109 595 something else that came earlier than language which involves the
rlm@109 596 ability to use your offline intelligence to discover something that
rlm@109 597 has a rich mathematical structure.
rlm@109 598 </p>
rlm@109 599 </div>
rlm@109 600
rlm@109 601 </div>
rlm@109 602
rlm@109 603 <div id="outline-container-3-4" class="outline-3">
rlm@109 604 <h3 id="sec-3-4"><a name="example-gap" id="example-gap"></a><span class="section-number-3">3.4</span> Example: Even toddlers use sophisticated geometric knowledge</h3>
rlm@109 605 <div class="outline-text-3" id="text-3-4">
rlm@109 606
rlm@109 607 <p>[19:44] I'll give you a simple example: if you look through a gap, you
rlm@109 608 can see something that's on the other side of the gap. Now, you
rlm@109 609 <i>might</i> see what you want to see, or you might see only part of it. If
rlm@109 610 you want to see more of it, which way would you move? Well, you could
rlm@109 611 either move <i>sideways</i>, and see through the gap&mdash;and see it roughly
rlm@109 612 the same amount but a different part of it [if it's a ????], or you
rlm@109 613 could move <i>towards</i> the gap and then your view will widen as you
rlm@109 614 approach the gap. Now, there's a bit of mathematics in there, insofar
rlm@109 615 as you are implicitly assuming that information travels in straight
rlm@109 616 lines, and as you go closer to a gap, the straight lines that you can
rlm@109 617 draw from where you are through the gap, widen as you approach that
rlm@109 618 gap. Now, there's a kind of theorem of Euclidean geometry in there
rlm@109 619 which I'm not going to try to state very precisely (and as far as I
rlm@109 620 know, wasn't stated explicitly in Euclidean geometry) but it's
rlm@109 621 something every toddler&mdash; human toddler&mdash;learns. (Maybe other
rlm@109 622 animals also know it, I don't know.) But there are many more things,
rlm@109 623 actions to perform, to get you more information about things, actions
rlm@109 624 to perform to conceal information from other people, actions that will
rlm@109 625 enable you to operate, to act on a rigid object in one place in order
rlm@109 626 to produce an effect on another place. So, there's a lot of stuff that
rlm@109 627 involves lines and rotations and angles and speeds and so on that I
rlm@109 628 think humans (maybe, to a lesser extent, other animals) develop the
rlm@109 629 ability to think about in a generic way. That means that you could
rlm@109 630 take out the generalizations from the particular contexts and then
rlm@109 631 re-use them in a new contexts in ways that I think are not yet
rlm@109 632 represented at all in AI and in theories of human learning in any []
rlm@109 633 way &mdash; although some people are trying to study learning of mathematics.
rlm@109 634 </p>
rlm@109 635 </div>
rlm@109 636 </div>
rlm@109 637
rlm@109 638 </div>
rlm@109 639
rlm@109 640 <div id="outline-container-4" class="outline-2">
rlm@109 641 <h2 id="sec-4"><span class="section-number-2">4</span> Animal intelligence</h2>
rlm@109 642 <div class="outline-text-2" id="text-4">
rlm@109 643
rlm@109 644
rlm@109 645
rlm@109 646 </div>
rlm@109 647
rlm@109 648 <div id="outline-container-4-1" class="outline-3">
rlm@109 649 <h3 id="sec-4-1"><span class="section-number-3">4.1</span> The priority is <i>cataloguing</i> what competences have evolved, not ranking them.</h3>
rlm@109 650 <div class="outline-text-3" id="text-4-1">
rlm@109 651
rlm@109 652 <p>[22:03] I wasn't going to challenge the claim that humans can do more
rlm@109 653 sophisticated forms of [tracking], just to mention that there are some
rlm@109 654 things that other animals can do which are in some ways comparable,
rlm@109 655 and some ways superior to [things] that humans can do. In particular,
rlm@109 656 there are species of birds and also, I think, some rodents ---
rlm@109 657 squirrels, or something &mdash; I don't know enough about the variety ---
rlm@109 658 that can hide nuts and remember where they've hidden them, and go back
rlm@109 659 to them. And there have been tests which show that some birds are able
rlm@109 660 to hide tens &mdash; you know, [eighteen] or something nuts &mdash; and to
rlm@109 661 remember which ones have been taken, which ones haven't, and so
rlm@109 662 on. And I suspect most humans can't do that. I wouldn't want to say
rlm@109 663 categorically that maybe we couldn't, because humans are very
rlm@109 664 [varied], and also [a few] people can develop particular competences
rlm@109 665 through training. But it's certainly not something I can do.
rlm@109 666 </p>
rlm@109 667
rlm@109 668 </div>
rlm@109 669
rlm@109 670 </div>
rlm@109 671
rlm@109 672 <div id="outline-container-4-2" class="outline-3">
rlm@109 673 <h3 id="sec-4-2"><span class="section-number-3">4.2</span> AI can be used to test philosophical theories</h3>
rlm@109 674 <div class="outline-text-3" id="text-4-2">
rlm@109 675
rlm@109 676 <p>[23:01] But I also would like to say that I am not myself particularly
rlm@109 677 interested in trying to align animal intelligences according to any
rlm@109 678 kind of scale of superiority; I'm just trying to understand what it
rlm@109 679 was that biological evolution produced, and how it works, and I'm
rlm@109 680 interested in AI <i>mainly</i> because I think that when one comes up with
rlm@109 681 theories about how these things work, one needs to have some way of
rlm@109 682 testing the theory. And AI provides ways of implementing and testing
rlm@109 683 theories that were not previously available: Immanuel Kant was trying
rlm@109 684 to come up with theories about how minds work, but he didn't have any
rlm@109 685 kind of a mechanism that he could build to test his theory about the
rlm@109 686 nature of mathematical knowledge, for instance, or how concepts were
rlm@109 687 developed from babyhood onward. Whereas now, if we do develop a
rlm@109 688 theory, we have a criterion of adequacy, namely it should be precise
rlm@109 689 enough and rich enough and detailed to enable a model to be
rlm@109 690 built. And then we can see if it works.
rlm@109 691 </p>
rlm@109 692 <p>
rlm@109 693 [24:07] If it works, it doesn't mean we've proved that the theory is
rlm@109 694 correct; it just shows it's a candidate. And if it doesn't work, then
rlm@109 695 it's not a candidate as it stands; it would need to be modified in
rlm@109 696 some way.
rlm@109 697 </p>
rlm@109 698 </div>
rlm@109 699 </div>
rlm@109 700
rlm@109 701 </div>
rlm@109 702
rlm@109 703 <div id="outline-container-5" class="outline-2">
rlm@109 704 <h2 id="sec-5"><span class="section-number-2">5</span> Is abstract general intelligence feasible?</h2>
rlm@109 705 <div class="outline-text-2" id="text-5">
rlm@109 706
rlm@109 707
rlm@109 708
rlm@109 709 </div>
rlm@109 710
rlm@109 711 <div id="outline-container-5-1" class="outline-3">
rlm@109 712 <h3 id="sec-5-1"><span class="section-number-3">5.1</span> It's misleading to compare the brain and its neurons to a computer made of transistors</h3>
rlm@109 713 <div class="outline-text-3" id="text-5-1">
rlm@109 714
rlm@109 715 <p>[24:27] I think there's a lot of optimism based on false clues:
rlm@109 716 the&hellip;for example, one of the false clues is to count the number of
rlm@109 717 neurons in the brain, and then talk about the number of transistors
rlm@109 718 you can fit into a computer or something, and then compare them. It
rlm@109 719 might turn out that the study of the way synapses work (which leads
rlm@109 720 some people to say that a typical synapse [] in the human brain has
rlm@109 721 computational power comparable to the Internet a few years ago,
rlm@109 722 because of the number of different molecules that are doing things,
rlm@109 723 the variety of types of things that are being done in those molecular
rlm@109 724 interactions, and the speed at which they happen, if you somehow count
rlm@109 725 up the number of operations per second or something, then you get
rlm@109 726 these comparable figures).
rlm@109 727 </p>
rlm@109 728 </div>
rlm@109 729
rlm@109 730 </div>
rlm@109 731
rlm@109 732 <div id="outline-container-5-2" class="outline-3">
rlm@109 733 <h3 id="sec-5-2"><span class="section-number-3">5.2</span> For example, brains may rely heavily on chemical information processing</h3>
rlm@109 734 <div class="outline-text-3" id="text-5-2">
rlm@109 735
rlm@109 736 <p>Now even if the details aren't right, there may just be a lot of
rlm@109 737 information processing that&hellip;going on in brains at the <i>molecular</i>
rlm@109 738 level, not the neural level. Then, if that's the case, the processing
rlm@109 739 units will be orders of magnitude larger in number than the number of
rlm@109 740 neurons. And it's certainly the case that all the original biological
rlm@109 741 forms of information processing were chemical; there weren't brains
rlm@109 742 around, and still aren't in most microbes. And even when humans grow
rlm@109 743 their brains, the process of starting from a fertilized egg and
rlm@109 744 producing this rich and complex structure is, for much of the time,
rlm@109 745 under the control of chemical computations, chemical information
rlm@109 746 processing&mdash;of course combined with physical sorts of materials and
rlm@109 747 energy and so on as well.
rlm@109 748 </p>
rlm@109 749 <p>
rlm@109 750 [26:25] So it would seem very strange if all that capability was
rlm@109 751 something thrown away when you've got a brain and all the information
rlm@109 752 processing, the [challenges that were handled in making a brain],
rlm@109 753 &hellip; This is handwaving on my part; I'm just saying that we <i>might</i>
rlm@109 754 learn that what brains do is not what we think they do, and that
rlm@109 755 problems of replicating them are not what we think they are, solely in
rlm@109 756 terms of numerical estimate of time scales, the number of components,
rlm@109 757 and so on.
rlm@109 758 </p>
rlm@109 759 </div>
rlm@109 760
rlm@109 761 </div>
rlm@109 762
rlm@109 763 <div id="outline-container-5-3" class="outline-3">
rlm@109 764 <h3 id="sec-5-3"><span class="section-number-3">5.3</span> Brain algorithms may simply be optimized for certain kinds of information processing other than bit manipulations</h3>
rlm@109 765 <div class="outline-text-3" id="text-5-3">
rlm@109 766
rlm@109 767 <p>[26:56] But apart from that, the other basis of skepticism concerns
rlm@109 768 how well we understand what the problems are. I think there are many
rlm@109 769 people who try to formalize the problems of designing an intelligent
rlm@109 770 system in terms of streams of information thought of as bit streams or
rlm@109 771 collections of bit streams, and they think of as the problems of
rlm@109 772 intelligence as being the construction or detection of patterns in
rlm@109 773 those, and perhaps not just detection of patterns, but detection of
rlm@109 774 patterns that are useable for sending <i>out</i> streams to control motors
rlm@109 775 and so on in order to []. And that way of conceptualizing the problem
rlm@109 776 may lead on the one hand to oversimplification, so that the things
rlm@109 777 that <i>would</i> be achieved, if those goals were achieved, maybe much
rlm@109 778 simpler, in some ways inadequate. Or the replication of human
rlm@109 779 intelligence, or the matching of human intelligence&mdash;or for that
rlm@109 780 matter, squirrel intelligence&mdash;but in another way, it may also make
rlm@109 781 the problem harder: it may be that some of the kinds of things that
rlm@109 782 biological evolution has achieved can't be done that way. And one of
rlm@109 783 the ways that might turn out to be the case is not because it's not
rlm@109 784 impossible in principle to do some of the information processing on
rlm@109 785 artificial computers-based-on-transistors and other bit-manipulating
rlm@109 786 []&mdash;but it may just be that the computational complexity of solving
rlm@109 787 problems, processes, or finding solutions to complex problems, are
rlm@109 788 much greater and therefore you might need a much larger universe than
rlm@109 789 we have available in order to do things.
rlm@109 790 </p>
rlm@109 791 </div>
rlm@109 792
rlm@109 793 </div>
rlm@109 794
rlm@109 795 <div id="outline-container-5-4" class="outline-3">
rlm@109 796 <h3 id="sec-5-4"><span class="section-number-3">5.4</span> Example: find the shortest path by dangling strings</h3>
rlm@109 797 <div class="outline-text-3" id="text-5-4">
rlm@109 798
rlm@109 799 <p>[28:55] Then if the underlying mechanisms were different, the
rlm@109 800 information processing mechanisms, they might be better tailored to
rlm@109 801 particular sorts of computation. There's a [] example, which is
rlm@109 802 finding the shortest route if you've got a collection of roads, and
rlm@109 803 they may be curved roads, and lots of tangled routes from A to B to C,
rlm@109 804 and so on. And if you start at A and you want to get to Z &mdash; a place
rlm@109 805 somewhere on that map &mdash; the process of finding the shortest route
rlm@109 806 will involve searching through all these different possibilities and
rlm@109 807 rejecting some that are longer than others and so on. But if you make
rlm@109 808 a model of that map out of string, where these strings are all laid
rlm@109 809 out on the maps and so have the lengths of the routes. Then if you
rlm@109 810 hold the two knots in the string &ndash; it's a network of string &mdash; which
rlm@109 811 correspond to the start point and end point, then <i>pull</i>, then the
rlm@109 812 bits of string that you're left with in a straight line will give you
rlm@109 813 the shortest route, and that process of pulling just gets you the
rlm@109 814 solution very rapidly in a parallel computation, where all the others
rlm@109 815 just hang by the wayside, so to speak.
rlm@109 816 </p>
rlm@109 817 </div>
rlm@109 818
rlm@109 819 </div>
rlm@109 820
rlm@109 821 <div id="outline-container-5-5" class="outline-3">
rlm@109 822 <h3 id="sec-5-5"><span class="section-number-3">5.5</span> In sum, we know surprisingly little about the kinds of problems that evolution solved, and the manner in which they were solved.</h3>
rlm@109 823 <div class="outline-text-3" id="text-5-5">
rlm@109 824
rlm@109 825 <p>[30:15] Now, I'm not saying brains can build networks of string and
rlm@109 826 pull them or anything like that; that's just an illustration of how if
rlm@109 827 you have the right representation, correctly implemented&mdash;or suitably
rlm@109 828 implemented&mdash;for a problem, then you can avoid very combinatorially
rlm@109 829 complex searches, which will maybe grow exponentially with the number
rlm@109 830 of components in your map, whereas with this thing, the time it takes
rlm@109 831 won't depend on how many strings you've [got on the map]; you just
rlm@109 832 pull, and it will depend only on the shortest route that exists in
rlm@109 833 there. Even if that shortest route wasn't obvious on the original map.
rlm@109 834 </p>
rlm@109 835
rlm@109 836 <p>
rlm@109 837 [30:59] So that's a rather long-winded way of formulating the
rlm@109 838 conjecture which&mdash;of supporting, a roundabout way of supporting the
rlm@109 839 conjecture that there may be something about the way molecules perform
rlm@109 840 computations where they have the combination of continuous change as
rlm@109 841 things move through space and come together and move apart, and
rlm@109 842 whatever &mdash; and also snap into states that then persist, so [as you
rlm@109 843 learn from] quantum mechanics, you can have stable molecular
rlm@109 844 structures which are quite hard to separate, and then in catalytic
rlm@109 845 processes you can separate them, or extreme temperatures, or strong
rlm@109 846 forces, but they may nevertheless be able to move very rapidly in some
rlm@109 847 conditions in order to perform computations.
rlm@109 848 </p>
rlm@109 849 <p>
rlm@109 850 [31:49] Now there may be things about that kind of structure that
rlm@109 851 enable searching for solutions to <i>certain</i> classes of problems to be
rlm@109 852 done much more efficiently (by brain) than anything we could do with
rlm@109 853 computers. It's just an open question.
rlm@109 854 </p>
rlm@109 855 <p>
rlm@109 856 [32:04] So it <i>might</i> turn out that we need new kinds of technology
rlm@109 857 that aren't on the horizon in order to replicate the functions that
rlm@109 858 animal brains perform &mdash;or, it might not. I just don't know. I'm not
rlm@109 859 claiming that there's strong evidence for that; I'm just saying that
rlm@109 860 it might turn out that way, partly because I think we know less than
rlm@109 861 many people think we know about what biological evolution achieved.
rlm@109 862 </p>
rlm@109 863 <p>
rlm@109 864 [32:28] There are some other possibilities: we may just find out that
rlm@109 865 there are shortcuts no one ever thought of, and it will all happen
rlm@109 866 much more quickly&mdash;I have an open mind; I'd be surprised, but it
rlm@109 867 could turn up. There <i>is</i> something that worries me much more than the
rlm@109 868 singularity that most people talk about, which is machines achieving
rlm@109 869 human-level intelligence and perhaps taking over [the] planet or
rlm@109 870 something. There's what I call the <i>singularity of cognitive catch-up</i> &hellip;
rlm@109 871 </p>
rlm@109 872 </div>
rlm@109 873 </div>
rlm@109 874
rlm@109 875 </div>
rlm@109 876
rlm@109 877 <div id="outline-container-6" class="outline-2">
rlm@109 878 <h2 id="sec-6"><span class="section-number-2">6</span> A singularity of cognitive catch-up</h2>
rlm@109 879 <div class="outline-text-2" id="text-6">
rlm@109 880
rlm@109 881
rlm@109 882
rlm@109 883 </div>
rlm@109 884
rlm@109 885 <div id="outline-container-6-1" class="outline-3">
rlm@109 886 <h3 id="sec-6-1"><span class="section-number-3">6.1</span> What if it will take a lifetime to learn enough to make something new?</h3>
rlm@109 887 <div class="outline-text-3" id="text-6-1">
rlm@109 888
rlm@109 889 <p>&hellip; SCC, singularity of cognitive catch-up, which I think we're close
rlm@109 890 to, or maybe have already reached&mdash;I'll explain what I mean by
rlm@109 891 that. One of the products of biological evolution&mdash;and this is one of
rlm@109 892 the answers to your earlier questions which I didn't get on to&mdash;is
rlm@109 893 that humans have not only the ability to make discoveries that none of
rlm@109 894 their ancestors have ever made, but to shorten the time required for
rlm@109 895 similar achievements to be reached by their offspring and their
rlm@109 896 descendants. So once we, for instance, worked out ways of complex
rlm@109 897 computations, or ways of building houses, or ways of finding our way
rlm@109 898 around, we don't need&hellip;our children don't need to work it out for
rlm@109 899 themselves by the same lengthy trial and error procedure; we can help
rlm@109 900 them get there much faster.
rlm@109 901 </p>
rlm@109 902 <p>
rlm@109 903 Okay, well, what I've been referring to as the singularity of
rlm@109 904 cognitive catch-up depends on the fact that&mdash;fairly obvious, and it's
rlm@109 905 often been commented on&mdash;that in case of humans, it's not necessary
rlm@109 906 for each generation to learn what previous generations learned <i>in the same way</i>. And we can speed up learning once something has been
rlm@109 907 learned, [it is able to] be learned by new people. And that has meant
rlm@109 908 that the social processes that support that kind of education of the
rlm@109 909 young can enormously accelerate what would have taken&hellip;perhaps
rlm@109 910 thousands [or] millions of years for evolution to produce, can happen in
rlm@109 911 a much shorter time.
rlm@109 912 </p>
rlm@109 913
rlm@109 914 <p>
rlm@109 915 [34:54] But here's the catch: in order for a new advance to happen ---
rlm@109 916 so for something new to be discovered that wasn't there before, like
rlm@109 917 Newtonian mechanics, or the theory of relativity, or Beethoven's music
rlm@109 918 or [style] or whatever &mdash; the individuals have to have traversed a
rlm@109 919 significant amount of what their ancestors have learned, even if they
rlm@109 920 do it much faster than their ancestors, to get to the point where they
rlm@109 921 can see the gaps, the possibilities for going further than their
rlm@109 922 ancestors, or their parents or whatever, have done.
rlm@109 923 </p>
rlm@109 924 <p>
rlm@109 925 [35:27] Now in the case of knowledge of science, mathematics,
rlm@109 926 philosophy, engineering and so on, there's been a lot of accumulated
rlm@109 927 knowledge. And humans are living a <i>bit</i> longer than they used to, but
rlm@109 928 they're still living for [whatever it is], a hundred years, or for
rlm@109 929 most people, less than that. So you can imagine that there might come
rlm@109 930 a time when in a normal human lifespan, it's not possible for anyone
rlm@109 931 to learn enough to understand the scope and limits of what's already
rlm@109 932 been achieved in order to see the potential for going beyond it and to
rlm@109 933 build on what's already been done to make that&hellip;those future steps.
rlm@109 934 </p>
rlm@109 935 <p>
rlm@109 936 [36:10] So if we reach that stage, we will have reached the
rlm@109 937 singularity of cognitive catch-up because the process of education
rlm@109 938 that enables individuals to learn faster than their ancestors did is
rlm@109 939 the catching-up process, and it may just be that we at some point
rlm@109 940 reach a point where catching up can only happen within a lifetime of
rlm@109 941 an individual, and after that they're dead and they can't go
rlm@109 942 beyond. And I have some evidence that there's a lot of that around
rlm@109 943 because I see a lot of people coming up with what <i>they</i> think of as
rlm@109 944 new ideas which they've struggled to come up with, but actually they
rlm@109 945 just haven't taken in some of what was&hellip;some of what was done [] by
rlm@109 946 other people, in other places before them. And I think that despite
rlm@109 947 the availability of search engines which make it <i>easier</i> for people
rlm@109 948 to get the information&mdash;for instance, when I was a student, if I
rlm@109 949 wanted to find out what other people had done in the field, it was a
rlm@109 950 laborious process&mdash;going to the library, getting books, and
rlm@109 951 &mdash;whereas now, I can often do things in seconds that would have taken
rlm@109 952 hours. So that means that if seconds [are needed] for that kind of
rlm@109 953 work, my lifespan has been extended by a factor of ten or
rlm@109 954 something. So maybe that <i>delays</i> the singularity, but it may not
rlm@109 955 delay it enough. But that's an open question; I don't know. And it may
rlm@109 956 just be that in some areas, this is more of a problem than others. For
rlm@109 957 instance, it may be that in some kinds of engineering, we're handing
rlm@109 958 over more and more of the work to machines anyways and they can go on
rlm@109 959 doing it. So for instance, most of the production of computers now is
rlm@109 960 done by a computer-controlled machine&mdash;although some of the design
rlm@109 961 work is done by humans&mdash; a lot of <i>detail</i> of the design is done by
rlm@109 962 computers, and they produce the next generation, which then produces
rlm@109 963 the next generation, and so on.
rlm@109 964 </p>
rlm@109 965 <p>
rlm@109 966 [37:57] I don't know if humans can go on having major advances, so
rlm@109 967 it'll be kind of sad if we can't.
rlm@109 968 </p>
rlm@109 969 </div>
rlm@109 970 </div>
rlm@109 971
rlm@109 972 </div>
rlm@109 973
rlm@109 974 <div id="outline-container-7" class="outline-2">
rlm@109 975 <h2 id="sec-7"><span class="section-number-2">7</span> Spatial reasoning: a difficult problem</h2>
rlm@109 976 <div class="outline-text-2" id="text-7">
rlm@109 977
rlm@109 978
rlm@109 979 <p>
rlm@109 980 [38:15] Okay, well, there are different problems [ ] mathematics, and
rlm@109 981 they have to do with properties. So for instance a lot of mathematics
rlm@109 982 that can be expressed in terms of logical structures or algebraic
rlm@109 983 structures and those are pretty well suited for manipulation and&hellip;on
rlm@109 984 computers, and if a problem can be specified using the
rlm@109 985 logical/algebraic notation, and the solution method requires creating
rlm@109 986 something in that sort of notation, then computers are pretty good,
rlm@109 987 and there are lots of mathematical tools around&mdash;there are theorem
rlm@109 988 provers and theorem checkers, and all kinds of things, which couldn't
rlm@109 989 have existed fifty, sixty years ago, and they will continue getting
rlm@109 990 better.
rlm@109 991 </p>
rlm@109 992
rlm@109 993 <p>
rlm@109 994 But there was something that I was <a href="#sec-3-4">alluding to earlier</a> when I gave the
rlm@109 995 example of how you can reason about what you will see by changing your
rlm@109 996 position in relation to a door, where what you are doing is using your
rlm@109 997 grasp of spatial structures and how as one spatial relationship
rlm@109 998 changes namely you come closer to the door or move sideways and
rlm@109 999 parallel to the wall or whatever, other spatial relationships change
rlm@109 1000 in parallel, so the lines from your eyes through to other parts of
rlm@109 1001 the&hellip;parts of the room on the other side of the doorway change,
rlm@109 1002 spread out more as you go towards the doorway, and as you move
rlm@109 1003 sideways, they don't spread out differently, but focus on different
rlm@109 1004 parts of the internal &hellip; that they access different parts of the
rlm@109 1005 &hellip; of the room.
rlm@109 1006 </p>
rlm@109 1007 <p>
rlm@109 1008 Now, those are examples of ways of thinking about relationships and
rlm@109 1009 changing relationships which are not the same as thinking about what
rlm@109 1010 happens if I replace this symbol with that symbol, or if I substitute
rlm@109 1011 this expression in that expression in a logical formula. And at the
rlm@109 1012 moment, I do not believe that there is anything in AI amongst the
rlm@109 1013 mathematical reasoning community, the theorem-proving community, that
rlm@109 1014 can model the processes that go on when a young child starts learning
rlm@109 1015 to do Euclidean geometry and is taught things about&mdash;for instance, I
rlm@109 1016 can give you a proof that the angles of any triangle add up to a
rlm@109 1017 straight line, 180 degrees.
rlm@109 1018 </p>
rlm@109 1019
rlm@109 1020 </div>
rlm@109 1021
rlm@109 1022 <div id="outline-container-7-1" class="outline-3">
rlm@109 1023 <h3 id="sec-7-1"><span class="section-number-3">7.1</span> Example: Spatial proof that the angles of any triangle add up to a half-circle</h3>
rlm@109 1024 <div class="outline-text-3" id="text-7-1">
rlm@109 1025
rlm@109 1026 <p>There are standard proofs which involves starting with one triangle,
rlm@109 1027 then adding a line parallel to the base one of my former students,
rlm@109 1028 Mary Pardoe, came up with which I will demonstrate with this &lt;he holds
rlm@109 1029 up a pen&gt; &mdash; can you see it? If I have a triangle here that's got
rlm@109 1030 three sides, if I put this thing on it, on one side &mdash; let's say the
rlm@109 1031 bottom&mdash;I can rotate it until it lies along the second&hellip;another
rlm@109 1032 side, and then maybe move it up to the other end ~. Then I can rotate
rlm@109 1033 it again, until it lies on the third side, and move it back to the
rlm@109 1034 other end. And then I'll rotate it again and it'll eventually end up
rlm@109 1035 on the original side, but it will have changed the direction it's
rlm@109 1036 pointing in &mdash; and it won't have crossed over itself so it will have
rlm@109 1037 gone through a half-circle, and that says that the three angles of a
rlm@109 1038 triangle add up to the rotations of half a circle, which is a
rlm@109 1039 beautiful kind of proof and almost anyone can understand it. Some
rlm@109 1040 mathematicians don't like it, because they say it hides some of the
rlm@109 1041 assumptions, but nevertheless, as far as I'm concerned, it's an
rlm@109 1042 example of a human ability to do reasoning which, once you've
rlm@109 1043 understood it, you can see will apply to any triangle &mdash; it's got to
rlm@109 1044 be a planar triangle &mdash; not a triangle on a globe, because then the
rlm@109 1045 angles can add up to more than &hellip; you can have three <i>right</i> angles
rlm@109 1046 if you have an equator&hellip;a line on the equator, and a line going up to
rlm@109 1047 to the north pole of the earth, and then you have a right angle and
rlm@109 1048 then another line going down to the equator, and you have a right
rlm@109 1049 angle, right angle, right angle, and they add up to more than a
rlm@109 1050 straight line. But that's because the triangle isn't in the plane,
rlm@109 1051 it's on a curved surface. In fact, that's one of the
rlm@109 1052 differences&hellip;definitional differences you can take between planar and
rlm@109 1053 curved surfaces: how much the angles of a triangle add up to. But our
rlm@109 1054 ability to <i>visualize</i> and notice the generality in that process, and
rlm@109 1055 see that you're going to be able to do the same thing using triangles
rlm@109 1056 that stretch in all sorts of ways, or if it's a million times as
rlm@109 1057 large, or if it's made&hellip;you know, written on, on&hellip;if it's drawn in
rlm@109 1058 different colors or whatever &mdash; none of that's going to make any
rlm@109 1059 difference to the essence of that process. And that ability to see
rlm@109 1060 the commonality in a spatial structure which enables you to draw some
rlm@109 1061 conclusions with complete certainty&mdash;subject to the possibility that
rlm@109 1062 sometimes you make mistakes, but when you make mistakes, you can
rlm@109 1063 discover them, as has happened in the history of geometrical theorem
rlm@109 1064 proving. Imre Lakatos had a wonderful book called <a href="http://en.wikipedia.org/wiki/Proofs_and_Refutations"><i>Proofs and Refutations</i></a> &mdash; which I won't try to summarize &mdash; but he has
rlm@109 1065 examples: mistakes were made; that was because people didn't always
rlm@109 1066 realize there were subtle subcases which had slightly different
rlm@109 1067 properties, and they didn't take account of that. But once they're
rlm@109 1068 noticed, you rectify that.
rlm@109 1069 </p>
rlm@109 1070 </div>
rlm@109 1071
rlm@109 1072 </div>
rlm@109 1073
rlm@109 1074 <div id="outline-container-7-2" class="outline-3">
rlm@109 1075 <h3 id="sec-7-2"><span class="section-number-3">7.2</span> Geometric results are fundamentally different than experimental results in chemistry or physics.</h3>
rlm@109 1076 <div class="outline-text-3" id="text-7-2">
rlm@109 1077
rlm@109 1078 <p>[43:28] But it's not the same as doing experiments in chemistry and
rlm@109 1079 physics, where you can't be sure it'll be the same on [] or at a high
rlm@109 1080 temperature, or in a very strong magnetic field &mdash; with geometric
rlm@109 1081 reasoning, in some sense you've got the full information in front of
rlm@109 1082 you; even if you don't always notice an important part of it. So, that
rlm@109 1083 kind of reasoning (as far as I know) is not implemented anywhere in a
rlm@109 1084 computer. And most people who do research on trying to model
rlm@109 1085 mathematical reasoning, don't pay any attention to that, because of
rlm@109 1086 &hellip; they just don't think about it. They start from somewhere else,
rlm@109 1087 maybe because of how they were educated. I was taught Euclidean
rlm@109 1088 geometry at school. Were you?
rlm@109 1089 </p>
rlm@109 1090 <p>
rlm@109 1091 (Adam ford: Yeah)
rlm@109 1092 </p>
rlm@109 1093 <p>
rlm@109 1094 Many people are not now. Instead they're taught set theory, and
rlm@109 1095 logic, and arithmetic, and [algebra], and so on. And so they don't use
rlm@109 1096 that bit of their brains, without which we wouldn't have built any of
rlm@109 1097 the cathedrals, and all sorts of things we now depend on.
rlm@109 1098 </p>
rlm@109 1099 </div>
rlm@109 1100 </div>
rlm@109 1101
rlm@109 1102 </div>
rlm@109 1103
rlm@109 1104 <div id="outline-container-8" class="outline-2">
rlm@109 1105 <h2 id="sec-8"><span class="section-number-2">8</span> Is near-term artificial general intelligence likely?</h2>
rlm@109 1106 <div class="outline-text-2" id="text-8">
rlm@109 1107
rlm@109 1108
rlm@109 1109
rlm@109 1110 </div>
rlm@109 1111
rlm@109 1112 <div id="outline-container-8-1" class="outline-3">
rlm@109 1113 <h3 id="sec-8-1"><span class="section-number-3">8.1</span> Two interpretations: a single mechanism for all problems, or many mechanisms unified in one program.</h3>
rlm@109 1114 <div class="outline-text-3" id="text-8-1">
rlm@109 1115
rlm@109 1116
rlm@109 1117 <p>
rlm@109 1118 [44:35] Well, this relates to what's meant by general. And when I
rlm@109 1119 first encountered the AGI community, I thought that what they all
rlm@109 1120 meant by general intelligence was <i>uniform</i> intelligence ---
rlm@109 1121 intelligence based on some common simple (maybe not so simple, but)
rlm@109 1122 single powerful mechanism or principle of inference. And there are
rlm@109 1123 some people in the community who are trying to produce things like
rlm@109 1124 that, often in connection with algorithmic information theory and
rlm@109 1125 computability of information, and so on. But there's another sense of
rlm@109 1126 general which means that the system of general intelligence can do
rlm@109 1127 lots of different things, like perceive things, understand language,
rlm@109 1128 move around, make things, and so on &mdash; perhaps even enjoy a joke;
rlm@109 1129 that's something that's not nearly on the horizon, as far as I
rlm@109 1130 know. Enjoying a joke isn't the same as being able to make laughing
rlm@109 1131 noises.
rlm@109 1132 </p>
rlm@109 1133 <p>
rlm@109 1134 Given, then, that there are these two notions of general
rlm@109 1135 intelligence&mdash;there's one that looks for one uniform, possibly
rlm@109 1136 simple, mechanism or collection of ideas and notations and algorithms,
rlm@109 1137 that will deal with any problem that's solvable &mdash; and the other
rlm@109 1138 that's general in the sense that it can do lots of different things
rlm@109 1139 that are combined into an integrated architecture (which raises lots
rlm@109 1140 of questions about how you combine these things and make them work
rlm@109 1141 together) and we humans, certainly, are of the second kind: we do all
rlm@109 1142 sorts of different things, and other animals also seem to be of the
rlm@109 1143 second kind, perhaps not as general as humans. Now, it may turn out
rlm@109 1144 that in some near future time, who knows&mdash;decades, a few
rlm@109 1145 decades&mdash;you'll be able to get machines that are capable of solving
rlm@109 1146 in a time that will depend on the nature of the problem, but any
rlm@109 1147 problem that is solvable, and they will be able to do it in some sort
rlm@109 1148 of tractable time &mdash; of course, there are some problems that are
rlm@109 1149 solvable that would require a larger universe and a longer history
rlm@109 1150 than the history of the universe, but apart from that constraint,
rlm@109 1151 these machines will be able to do anything []. But to be able to do
rlm@109 1152 some of the kinds of things that humans can do, like the kinds of
rlm@109 1153 geometrical reasoning where you look at the shape and you abstract
rlm@109 1154 away from the precise angles and sizes and shapes and so on, and
rlm@109 1155 realize there's something general here, as must have happened when our
rlm@109 1156 ancestors first made the discoveries that eventually put together in
rlm@109 1157 Euclidean geometry.
rlm@109 1158 </p>
rlm@109 1159 <p>
rlm@109 1160 It may be that that requires mechanisms of a kind that we don't know
rlm@109 1161 anything about at the moment. Maybe brains are using molecules and
rlm@109 1162 rearranging molecules in some way that supports that kind of
rlm@109 1163 reasoning. I'm not saying they are &mdash; I don't know, I just don't see
rlm@109 1164 any simple&hellip;any obvious way to map that kind of reasoning capability
rlm@109 1165 onto what we currently do on computers. There is&mdash;and I just
rlm@109 1166 mentioned this briefly beforehand&mdash;there is a kind of thing that's
rlm@109 1167 sometimes thought of as a major step in that direction, namely you can
rlm@109 1168 build a machine (or a software system) that can represent some
rlm@109 1169 geometrical structure, and then be told about some change that's going
rlm@109 1170 to happen to it, and it can predict in great detail what'll
rlm@109 1171 happen. And this happens for instance in game engines, where you say
rlm@109 1172 we have all these blocks on the table and I'll drop one other block,
rlm@109 1173 and then [the thing] uses Newton's laws and properties of rigidity of
rlm@109 1174 the parts and the elasticity and also stuff about geometries and space
rlm@109 1175 and so on, to give you a very accurate representation of what'll
rlm@109 1176 happen when this brick lands on this pile of things, [it'll bounce and
rlm@109 1177 go off, and so on]. And you just, with more memory and more CPU power,
rlm@109 1178 you can increase the accuracy&mdash; but that's totally different than
rlm@109 1179 looking at <i>one</i> example, and working out what will happen in a whole
rlm@109 1180 <i>range</i> of cases at a higher level of abstraction, whereas the game
rlm@109 1181 engine does it in great detail for <i>just</i> this case, with <i>just</i> those
rlm@109 1182 precise things, and it won't even know what the generalizations are
rlm@109 1183 that it's using that would apply to others []. So, in that sense, [we]
rlm@109 1184 may get AGI &mdash; artificial general intelligence &mdash; pretty soon, but
rlm@109 1185 it'll be limited in what it can do. And the other kind of general
rlm@109 1186 intelligence which combines all sorts of different things, including
rlm@109 1187 human spatial geometrical reasoning, and maybe other things, like the
rlm@109 1188 ability to find things funny, and to appreciate artistic features and
rlm@109 1189 other things may need forms of pattern-mechanism, and I have an open
rlm@109 1190 mind about that.
rlm@109 1191 </p>
rlm@109 1192 </div>
rlm@109 1193 </div>
rlm@109 1194
rlm@109 1195 </div>
rlm@109 1196
rlm@109 1197 <div id="outline-container-9" class="outline-2">
rlm@109 1198 <h2 id="sec-9"><span class="section-number-2">9</span> Abstract General Intelligence impacts</h2>
rlm@109 1199 <div class="outline-text-2" id="text-9">
rlm@109 1200
rlm@109 1201
rlm@109 1202 <p>
rlm@109 1203 [49:53] Well, as far as the first type's concerned, it could be useful
rlm@109 1204 for all kinds of applications &mdash; there are people who worry about
rlm@109 1205 where there's a system that has that type of intelligence, might in
rlm@109 1206 some sense take over control of the planet. Well, humans often do
rlm@109 1207 stupid things, and they might do something stupid that would lead to
rlm@109 1208 disaster, but I think it's more likely that there would be other
rlm@109 1209 things [] lead to disaster&mdash; population problems, using up all the
rlm@109 1210 resources, destroying ecosystems, and whatever. But certainly it would
rlm@109 1211 go on being useful to have these calculating devices. Now, as for the
rlm@109 1212 second kind of them, I don't know&mdash;if we succeeded at putting
rlm@109 1213 together all the parts that we find in humans, we might just make an
rlm@109 1214 artificial human, and then we might have some of them as your friends,
rlm@109 1215 and some of them we might not like, and some of them might become
rlm@109 1216 teachers or whatever, composers &mdash; but that raises a question: could
rlm@109 1217 they, in some sense, be superior to us, in their learning
rlm@109 1218 capabilities, their understanding of human nature, or maybe their
rlm@109 1219 wickedness or whatever &mdash; these are all issues in which I expect the
rlm@109 1220 best science fiction writers would give better answers than anything I
rlm@109 1221 could do, but I did once fantasize when I [back] in 1978, that perhaps
rlm@109 1222 if we achieved that kind of thing, that they would be wise, and gentle
rlm@109 1223 and kind, and realize that humans are an inferior species that, you
rlm@109 1224 know, have some good features, so they'd keep us in some kind of
rlm@109 1225 secluded&hellip;restrictive kind of environment, keep us away from
rlm@109 1226 dangerous weapons, and so on. And find ways of cohabitating with
rlm@109 1227 us. But that's just fantasy.
rlm@109 1228 </p>
rlm@109 1229 <p>
rlm@109 1230 Adam Ford: Awesome. Yeah, there's an interesting story <i>With Folded Hands</i> where [the computers] want to take care of us and want to
rlm@109 1231 reduce suffering and end up lobotomizing everybody [but] keeping them
rlm@109 1232 alive so as to reduce the suffering.
rlm@109 1233 </p>
rlm@109 1234 <p>
rlm@109 1235 Aaron Sloman: Not all that different from <i>Brave New World</i>, where it
rlm@109 1236 was done with drugs and so on, but different humans are given
rlm@109 1237 different roles in that system, yeah.
rlm@109 1238 </p>
rlm@109 1239 <p>
rlm@109 1240 There's also <i>The Time Machine</i>, H.G. Wells, where the &hellip; in the
rlm@109 1241 distant future, humans have split in two: the Eloi, I think they were
rlm@109 1242 called, they lived underground, they were the [] ones, and then&mdash;no,
rlm@109 1243 the Morlocks lived underground; Eloi lived on the planet; they were
rlm@109 1244 pleasant and pretty but not very bright, and so on, and they were fed
rlm@109 1245 on by &hellip;
rlm@109 1246 </p>
rlm@109 1247 <p>
rlm@109 1248 Adam Ford: [] in the future.
rlm@109 1249 </p>
rlm@109 1250 <p>
rlm@109 1251 Aaron Sloman: As I was saying, if you ask science fiction writers,
rlm@109 1252 you'll probably come up with a wide variety of interesting answers.
rlm@109 1253 </p>
rlm@109 1254 <p>
rlm@109 1255 Adam Ford: I certainly have; I've spoken to [] of Birmingham, and
rlm@109 1256 Sean Williams, &hellip; who else?
rlm@109 1257 </p>
rlm@109 1258 <p>
rlm@109 1259 Aaron Sloman: Did you ever read a story by E.M. Forrester called <i>The Machine Stops</i> &mdash; very short story, it's <a href="http://archive.ncsa.illinois.edu/prajlich/forster.html">on the Internet somewhere</a>
rlm@109 1260 &mdash; it's about a time when people sitting &hellip; and this was written in
rlm@109 1261 about [1914 ] so it's about&hellip;over a hundred years ago &hellip; people are
rlm@109 1262 in their rooms, they sit in front of screens, and they type things,
rlm@109 1263 and they communicate with one another that way, and they don't meet;
rlm@109 1264 they have debates, and they give lectures to their audiences that way,
rlm@109 1265 and then there's a woman whose son says &ldquo;I'd like to see
rlm@109 1266 you&rdquo; and she says &ldquo;What's the point? You've got me at
rlm@109 1267 this point &rdquo; but he wants to come and talk to her &mdash; I won't
rlm@109 1268 tell you how it ends, but.
rlm@109 1269 </p>
rlm@109 1270 <p>
rlm@109 1271 Adam Ford: Reminds me of the Internet.
rlm@109 1272 </p>
rlm@109 1273 <p>
rlm@109 1274 Aaron Sloman: Well, yes; he invented &hellip; it was just extraordinary
rlm@109 1275 that he was able to do that, before most of the components that we
rlm@109 1276 need for it existed.
rlm@109 1277 </p>
rlm@109 1278 <p>
rlm@109 1279 Adam Ford: [Another person who did that] was Vernor Vinge [] <i>True Names</i>.
rlm@109 1280 </p>
rlm@109 1281 <p>
rlm@109 1282 Aaron Sloman: When was that written?
rlm@109 1283 </p>
rlm@109 1284 <p>
rlm@109 1285 Adam Ford: The seventies.
rlm@109 1286 </p>
rlm@109 1287 <p>
rlm@109 1288 Aaron Sloman: Okay, well a lot of the technology was already around
rlm@109 1289 then. The original bits of internet were working, in about 1973, I was
rlm@109 1290 sitting &hellip; 1974, I was sitting at Sussex University trying to
rlm@109 1291 use&hellip;learn LOGO, the programming language, to decide whether it was
rlm@109 1292 going to be useful for teaching AI, and I was sitting [] paper
rlm@109 1293 teletype, there was paper coming out, transmitting ten characters a
rlm@109 1294 second from Sussex to UCL computer lab by telegraph cable, from there
rlm@109 1295 to somewhere in Norway via another cable, from there by satellite to
rlm@109 1296 California to a computer Xerox [] research center where they had
rlm@109 1297 implemented a computer with a LOGO system on it, with someone I had
rlm@109 1298 met previously in Edinburgh, Danny Bobrow, and he allowed me to have
rlm@109 1299 access to this sytem. So there I was typing. And furthermore, it was
rlm@109 1300 duplex typing, so every character I typed didn't show up on my
rlm@109 1301 terminal until it had gone all the way there and echoed back, so I
rlm@109 1302 would type, and the characters would come back four seconds later.
rlm@109 1303 </p>
rlm@109 1304 <p>
rlm@109 1305 [55:26] But that was the Internet, and I think Vernor Vinge was
rlm@109 1306 writing after that kind of thing had already started, but I don't
rlm@109 1307 know. Anyway.
rlm@109 1308 </p>
rlm@109 1309 <p>
rlm@109 1310 [55:41] Another&hellip;I mentioned H.G. Wells, <i>The Time Machine</i>. I
rlm@109 1311 recently discovered, because <a href="http://en.wikipedia.org/wiki/David_Lodge_(author)">David Lodge</a> had written a sort of
rlm@109 1312 semi-novel about him, that he had invented Wikipedia, in advance&mdash; he
rlm@109 1313 had this notion of an encyclopedia that was free to everybody, and
rlm@109 1314 everybody could contribute and [collaborate on it]. So, go to the
rlm@109 1315 science fiction writers to find out the future &mdash; well, a range of
rlm@109 1316 possible futures.
rlm@109 1317 </p>
rlm@109 1318 <p>
rlm@109 1319 Adam Ford: Well the thing is with science fiction writers, they have
rlm@109 1320 to maintain some sort of interest for their readers, after all the
rlm@109 1321 science fiction which reaches us is the stuff that publishers want to
rlm@109 1322 sell, and so there's a little bit of a &hellip; a bias towards making a
rlm@109 1323 plot device there, and so the dramatic sort of appeals to our
rlm@109 1324 amygdala, our lizard brain; we'll sort of stay there obviously to some
rlm@109 1325 extent. But I think that they do come up with sort of amazing ideas; I
rlm@109 1326 think it's worth trying to make these predictions; I think that we
rlm@109 1327 should more time on strategic forecasting, I mean take that seriously.
rlm@109 1328 </p>
rlm@109 1329 <p>
rlm@109 1330 Aaron Sloman: Well, I'm happy to leave that to others; I just want to
rlm@109 1331 try to understand these problems that bother me about how things
rlm@109 1332 work. And it may be that some would say that's irresponsible if I
rlm@109 1333 don't think about what the implications will be. Well, understanding
rlm@109 1334 how humans work <i>might</i> enable us to make [] humans &mdash; I suspect it
rlm@109 1335 wont happen in this century; I think it's going to be too difficult.
rlm@109 1336 </p></div>
rlm@109 1337 </div>
rlm@109 1338 </div>
rlm@109 1339
rlm@109 1340 <div id="postamble">
rlm@109 1341 <p class="date">Date: 2013-10-04 18:49:53 UTC</p>
rlm@109 1342 <p class="author">Author: Dylan Holmes</p>
rlm@109 1343 <p class="creator">Org version 7.7 with Emacs version 23</p>
rlm@109 1344 <a href="http://validator.w3.org/check?uri=referer">Validate XHTML 1.0</a>
rlm@109 1345
rlm@109 1346 </div>
rlm@109 1347 </body>
rlm@109 1348 </html>