{"id":32028,"date":"2018-11-13T15:38:34","date_gmt":"2018-11-13T15:38:34","guid":{"rendered":"http:\/\/rafaelfajardo.com\/portfolio\/common-sense-the-chomskypiaget-debates-come-to\/"},"modified":"2018-11-13T15:38:34","modified_gmt":"2018-11-13T15:38:34","slug":"common-sense-the-chomskypiaget-debates-come-to","status":"publish","type":"post","link":"https:\/\/rafaelfajardo.com\/portfolio\/common-sense-the-chomskypiaget-debates-come-to\/","title":{"rendered":"Common sense: the Chomsky\/Piaget debates come to AI"},"content":{"rendered":"

mostlysignssomeportents<\/a>:<\/p>\n

\n
<\/figure>\n

In 1975, Noam Chomsky and Jean Paiget held a historic debate about the
\nnature of human cognition; Chomsky held that babies are born with a
\nbunch of in-built rules and instincts that help them build up the
\nknowledge that they need to navigate the world; Piaget argued that
\nbabies are effectively blank slates that acquire knowledge from
\nexperiencing the world (including the knowledge that there is a thing
\ncalled \u201cexperience\u201d and \u201cthe world\u201d).<\/p>\n

\nFor most of AI\u2019s history, Chomsky\u2019s approach prevailed: computer
\nscientists painstakingly tried to equip computers with a baseline of
\nknowledge about the relationships between things in the world, hoping
\nthat computers would some day build up from this base to construct
\ncomplex, powerful reasoning systems.\n<\/p>\n

\nThe current machine learning revolution can be traced to a jettisoning
\nof this approach in favor of a Piaget-style blank slate, where layers of
\n neural nets are trained on massive corpuses of data (sometimes labeled
\nby hand, but often completely blank) and use equally massive computation
\n to make sense of the data, creating their own understanding of the
\nworld.\n<\/p>\n

\nPiaget-style deep learning has taken AI a long way in a short time, but it\u2019s hitting a wall. It\u2019s not just the weird and vastly entertaining local optima<\/a>
\n that these systems get stuck in: it\u2019s the huge corpuses of data needed
\nto train them and the inability of machine learning to generalize one
\nmodel to bootstrap another and another.\n<\/p>\n

\nThe fall-off the rate of progress in machine learning, combined with the
\n excitement that ML\u2019s recent gains provoked, has breathed new life into
\nthe Chomskyian approach to ML, and computer scientists all over the
\nworld are trying to create \u201ccommon sense\u201d corpuses of knowledge that
\nthey can imbue machine learning systems with before they are exposed to
\ntraining data.\n<\/p>\n

\nThis approach seems to be hurdling some of the walls that stopped the
\nPiaget-style ML. Some Chomskyian ML models attained a high degree of
\nefficiency with much smaller training data sets.\n<\/p>\n

\nFrequent Boing Boing contributor<\/a>
\n Clive Thompson\u2019s long piece on the state of the Chomsky\/Piaget debate
\nin ML is an excellent read, and really comes to the (retrospectively)
\nobvious conclusion: it doesn\u2019t really matter whether Chomsky or Piaget
\nare right about how kids learn, because each of them is right about how
\ncomputers learn \u2013 a little from Column A, a little from Column B.<\/p>\n

https:\/\/boingboing.net\/2018\/11\/13\/naive-learning.html<\/a><\/i><\/p>\n<\/blockquote>\n

at it\u2019s core AI and Machine Learning are hypotheses about how humans learn. That is to say they are guesses. Educated guesses by erudite people, but guesses all the same.<\/p>\n","protected":false},"excerpt":{"rendered":"

mostlysignssomeportents: In 1975, Noam Chomsky and Jean Paiget held a historic debate about the nature of human cognition; Chomsky held that babies are born with a bunch of in-built rules and instincts that help them build up the knowledge that they need to navigate the world; Piaget argued that babies are effectively blank slates that […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[],"tags":[1539],"class_list":["post-32028","post","type-post","status-publish","format-standard","hentry","tag-emergent-digital-practices"],"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_shortlink":"https:\/\/wp.me\/p6PWot-8kA","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/posts\/32028","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/comments?post=32028"}],"version-history":[{"count":0,"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/posts\/32028\/revisions"}],"wp:attachment":[{"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/media?parent=32028"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/categories?post=32028"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/rafaelfajardo.com\/portfolio\/wp-json\/wp\/v2\/tags?post=32028"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}