AI

Fortean Times - - Reviews/Books - Mark Greener

Its Na­ture and Fu­ture Mar­garet A Bo­den Oxford Univer­sity Press 2016 Hb, 198pp, ilus, ind, £12.99, ISBN 9780198777984

Op­ti­mists be­lieve that ar­ti­fi­cial in­tel­li­gence (AI) will help over­come al­most ev­ery is­sue fac­ing hu­man­ity, in­clud­ing war, pesti­lence, hunger and even, by up­load­ing con­scious­ness, death. Pes­simists en­vis­age a fu­ture closer to Skynet where the ma­chines take con­trol. I’m scep­ti­cal: AI, at least in our life­times, is prov­ing to be nei­ther as good nor as bad as op­ti­mists and pes­simists pre­dict. In­deed, I have a sneak­ing sus­pi­cion that if AI ever reaches the kitchen we’ll drink, as Arthur Dent found, “a liq­uid that was al­most, but not quite, en­tirely un­like tea”.

One of the prob­lems fac­ing any dis­cus­sion of AI is defin­ing what ‘in­tel­li­gence’ ac­tu­ally is: it’s a no­to­ri­ously sub­tle, mul­ti­fac­eted and enig­matic con­cept. I love pon­der­ing a chess prob­lem. Yet I can set my Fritz chess pro­gram to beat me eas­ily. My so­cial and emo­tional in­tel­li­gence would get me ex­pelled from the Amal­ga­mated Union of Wallflow­ers, Recluses and As­so­ci­ated Timid Peo­ple for be­ing too shy. Yet AI is be­gin­ning to model emo­tional in­tel­li­gence, such as when ‘com­puter com­pan­ions’ re­spond with sym­pa­thy or sex­u­ally al­lur­ing be­hav­iours and speech. Sooner or later they’ll beat me there as well, I sus­pect. Af­ter all, they’re all facets of in­tel­li­gence, all de­pend on pro­cess­ing data, all rely on eval­u­at­ing in­for­ma­tion. But are they re­ally more ‘in­tel­li­gent’ in the sense we gen­er­ally use the term? I hope not.

Bo­den de­fines AI as seek­ing “to make com­put­ers do the sorts of things that minds can do”. She points out that in­tel­li­gence is “a richly struc­tured space of di­verse in­for­ma­tion-pro­cess­ing ca­pac­i­ties”. This def­i­ni­tion al­lows Bo­den to cover the five ma­jor types of AI: clas­si­cal AI; ar­ti­fi­cial neu­ral net­works; evo­lu­tion­ary pro­gram­ming; cel­lu­lar au­tom­ata; and dy­nam­i­cal sys­tems. While the forms and uses of AI dif­fer widely, all are es­sen­tially sys­tems that process in­for­ma­tion.

Of­ten por­trayed as the wave of the fu­ture, many as­pects of AI have a sur­pris­ingly long in­tel­lec­tual her­itage. In the 1840s, for ex­am­ple, Lady Ada Lovelace pre­dicted el­e­ments that now form part of the foun­da­tions of AI – such as pro­cess­ing sym­bols that po­ten­tially rep­re­sent “all sub­jects in the uni­verse”. Lovelace’s in­ter­est in logic in­spired her de­scrip­tion of sev­eral ba­sic pro­gram­ming con­cepts, in­clud­ing stored pro­grams, hi­er­ar­chi­cal sub­rou­tines and bugs. In the late 1950s, Arthur Sa­muel de­vel­oped a pro­gram that beat its creator at draughts.

By the 1960s “an in­tel­lec­tual schism” had de­vel­oped be­tween AI re­searchers in­ter­ested in life, who worked in cy­ber­net­ics, and those in­ter­ested in mind, who worked on sym­bolic com­put­ing. Re­searchers in­ter­ested in net­works cov­ered mind and brain. But as they mainly stud­ied as­so­cia­tive learn­ing, they were closer to cy­ber­net­ics. Bo­den notes that “there was scant mu­tual re­spect be­tween th­ese in­creas­ingly sep­a­rate sub­groups”. She elo­quently traces the im­pli­ca­tions and de­vel­op­ments that arose from this schism. While AI dis­cus­sions in­evitably fo­cus on the fu­ture, Bo­den traces the dis­ci­pline’s fas­ci­nat­ing his­tory, which helps place all the hype in con­text.

We might not have the T-800, Marvin or Deep Thought. But AI drives robots on Mars, an­i­mates Hol­ly­wood movies, dis­tracts you with mo­bile phone apps, hope­fully gets you to where you’re go­ing with sat-navs, and pre­dicts stock mar­ket move­ments. AI is al­ready so ubiq­ui­tous that, ar­guably, we all need at least a pass­ing ac­quain­tance with its core con­cepts, ideas and trends. Bo­den’s book is an ex­cel­lent, ac­ces­si­ble in­tro­duc­tion even for the com­plete AI novice.

Less pro­saically, Bo­den notes that philoso­phers use AI con­cepts to help il­lu­mi­nate is­sues such as free will, con­scious­ness and hu­man cre­ativ­ity. Bi­ol­o­gists use AI to model as­pects of liv­ing or­gan­isms and hope­fully bet­ter un­der­stand the elu­sive na­ture of life. In­deed, de­spite AI’s achieve­ments, Bo­den’s elo­quent book also shows just how re­mark­able the hu­man brain re­ally is. “AI has taught us that hu­man minds are hugely richer, and more sub­tle, than psy­chol­o­gists pre­vi­ously imag­ined,” she writes. “In­deed that is the main les­son to be learned from AI” (ital­ics in orig­i­nal).

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.