{"id":1911,"date":"2023-05-24T15:43:22","date_gmt":"2023-05-24T14:43:22","guid":{"rendered":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/?p=1911"},"modified":"2023-05-24T16:06:10","modified_gmt":"2023-05-24T15:06:10","slug":"monochrome-mary-gpt-and-me","status":"publish","type":"post","link":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/2023\/05\/24\/monochrome-mary-gpt-and-me\/","title":{"rendered":"Monochrome Mary, GPT, and Me"},"content":{"rendered":"\n

Frank Jackson\u2019s hypothetical black-and-white room-reared color neuroscientist \u201cMary\u201d<\/a>, unlike ChatGPT, sees (but only black\/white\/gray). So her vocabulary<\/em> for setting up her explanation of color perception is nevertheless grounded<\/a>. (Even Helen Keller<\/a>\u2019s was, because, lacking sight and hearing since 19 months old, she still had touch and kinesthesia.) ChatGPT has no grounding at all.<\/p>\n\n\n\n

So we can get from \u201chorse, horn, vanish, eyes, observing instruments\u201d to the unobservable \u201cpeekaboo unicorn<\/a>\u201d purely verbally, using just those prior (grounded) words to define or describe it.<\/p>\n\n\n\n

But ChatGPT can\u2019t get anywhere, because it has no grounded vocabulary at all. Just words \u201cin context\u201d (\u201c100 trillion parameters and 300 billion words\u2026: 570 gigabytes of text data\u201d).<\/p>\n\n\n\n

So much for ChatGPT\u2019s chances of the \u201cphase transition\u201d some emergence enthusiasts are imagining — from just harvesting and organizing the words in ChatGPT\u2019s\u00a0\u00a0bloated 570gig belly, to an \u201cemergent\u201d\u00a0understanding<\/em>\u00a0of them. You, me, Helen Keller or Monochrome Mary, we could, in contrast, understand ChatGPT\u2019s words. And we could\u00a0mean<\/em>\u00a0them if we said them. \u00a0So could the dead authors of the words in ChatGPT\u2019s belly, once. That\u2019s the difference between (intrinsic) understanding and word-crunching by a well-fed statistical parrot.<\/p>\n\n\n\n

Two important, connected, details: (1) Mary would be more than surprised if her screen went from B\/W to color: she would become able to DO things that she could not do before (on her own) locked in her B\/W room — like tell apart red and green, just as would any daltonian, if their trichromacy were repaired.\u00a0<\/p>\n\n\n\n

More important, (2) Mary, or Helen, or any daltonianian could be told any number of words about what green is neurologically and what things are green and what not. But they cannot be told what seeing green FEELS-LIKE (i.e., what green looks-like). And that\u2019s the point. If it weren\u2019t for the \u201chard problem<\/a>\u201d \u2013 the fact that it\u00a0feels-like something<\/em>\u00a0to see, hear, touch [and\u00a0understand<\/em>\u00a0and\u00a0mean<\/em>] something (none of which is something you DO) \u2013 Mary, Helen, and a daltonian would not be missing anything about green. Grounding in transducer\/effector know-how alone would be all there was to \u201cmeaning\u201d and \u201cunderstanding.\u201d But it\u2019s not.\u00a0<\/p>\n\n\n\n

Mary, Helen and the daltonian can make do with (imperfect) verbal analogies and extrapolations from the senses they do have to the senses they lack. But ChatGPT can\u2019t. Because all it has is a bellyful of ungrounded words (but\u00a0syntactically structured<\/a>\u00a0by their grounded authors \u2013 some of whom are perhaps already underground, long dead and buried\u2026).<\/p>\n\n\n\n

So meaning and understanding is not just sensorimotor know-how. It\u2019s based on\u00a0sentience<\/em> too (in fact, first and foremost, though, so far, it\u2019s an unsolved hard-problem to explain how or why). ChatGPT (attention \u201csingularity\u201d fans!) lacks that completely.\u00a0<\/p>\n\n\n\n

It FEELS-LIKE something to understand <\/em>what red (or round, or anything) means<\/em>. And that\u2019s not just transducer\/effector know-how that percolates up from the sensorimotor grounding. It\u2019s what it feels-like to see, hear, touch, taste and manipulate all those things out there that we are talking about, with our words. <\/p>\n\n\n\n

ChatGPT is talking, but ABOUT nothing. The \u201caboutness\u201d is supplied by the original author\u2019s and the reader\u2019s head. ChatGPT is just producing  recombinatory, fill-in-the-blanks echolalia.<\/p>\n\n\n\n

\"\"
Monochrome Rainbow by DALL-E<\/figcaption><\/figure>\n\n\n\n

<\/p>\n","protected":false},"excerpt":{"rendered":"

Frank Jackson\u2019s hypothetical black-and-white room-reared color neuroscientist \u201cMary\u201d, unlike ChatGPT, sees (but only black\/white\/gray). So her vocabulary for setting up her explanation of color perception is nevertheless grounded. (Even Helen Keller\u2019s was, because, lacking sight and hearing since 19 months old, she still had touch and kinesthesia.) ChatGPT has no grounding at all. So we can get from \u201chorse, horn, … <\/p>\n

Continue reading “Monochrome Mary, GPT, and Me”<\/span><\/a><\/p>\n","protected":false},"author":3074,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[152,146,119],"tags":[],"class_list":["post-1911","post","type-post","status-publish","format-standard","hentry","category-hard-problem","category-chatgpt","category-sentience"],"jetpack_featured_media_url":"","_links":{"self":[{"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/posts\/1911","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/users\/3074"}],"replies":[{"embeddable":true,"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/comments?post=1911"}],"version-history":[{"count":5,"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/posts\/1911\/revisions"}],"predecessor-version":[{"id":1918,"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/posts\/1911\/revisions\/1918"}],"wp:attachment":[{"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/media?parent=1911"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/categories?post=1911"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/generic.wordpress.soton.ac.uk\/skywritings\/wp-json\/wp\/v2\/tags?post=1911"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}