MIND CON­TROL

Com­put­ers that are smarter than hu­mans? It’s closer than you think

Emirates Man - - CONTENTS -

Johnny Depp’s Tran­scen­dence imag­ines a world of hy­per in­tel­li­gent com­put­ing that out­strips the hu­man brain’s think­ing prow­ess. Paddy Smith ex­plains why it’s nei­ther far-fetched, nor far away

Switch your mind to ‘open’ and upload this mod­icum of data: com­put­ers could be more in­tel­li­gent than us within 30 years. They can al­ready beat us at chess (in 1997 IBM’s Deep Blue fa­mously clob­bered world cham­pion Garry Kas­parov) and out­class us at com­plex string cal­cu­la­tions (shortly af­ter the turn of the century com­put­ers helped us un­ravel our en­tire DNA se­quence). Now they can drive cars and tell you if you’re smil­ing, yet the most pow­er­ful data pro­cess­ing unit we know re­mains the hu­man brain. But for how long?

Luck­ily, we’ve had plenty of time to get used to the idea of in­vent­ing our way into sec­ond place. It’s nearly 200 years since Mary Shel­ley dreamt up a mad sci­en­tist called Vic­tor Franken­stein, whose cre­ation learns to speak and read be­fore de­mand­ing a fe­male coun­ter­part be man­u­fac­tured to keep him com­pany. Since Franken­stein, the hy­po­thet­i­cal su­pe­ri­or­ity of syn­thetic in­tel­li­gence has played sec­ond place only to alien life in the sci­ence- ction canon’s pre­ferred plots.

A few mod­ern nota­bles: Hal in Stan­ley Kubrick’s 2001: A Space Odyssey (1968); Ash in Ri­d­ley Scott’s Alien (1979); GERTY in Dun­can Jones’ Moon (2009). All these ctional ex­am­ples have some­thing in com­mon – like Dr Franken­stein’s monster, they turn against their cre­ators: us.

What none of them does is pro­vide the back­story for this pos­si­bil­ity and that is where Tran­scen­dence, the di­rec­to­rial de­but from Wally Pfis­ter star­ring Johnny Depp, takes up the slack. Depp plays Dr Will Caster, an ar­ti­fi­cial in­tel­li­gence (AI) evan­ge­list who gets shot but has his brain up­loaded to a com­puter in an at­tempt to sal­vage his ex­pan­sive knowl­edge. The sci­en­tist then be­comes the com­puter (or the com­puter he, com­plete with emo­tional pro­cess­ing) and be­comes mega­lo­ma­ni­a­cal, and so on.

Scared? No, me nei­ther. We’ve all seen it be­fore. But the the­ory is sound and be­com­ing closer to a prac­ti­cal re­al­ity daily. The com­put­ers that drive cars and beat you at chess will one day be able to learn bet­ter and faster than you. They won’t need to sleep. In­stead they will sink their en­ergy into de­vel­op­ing ever more pow­er­ful data-cruch­ing off­spring who, in turn, will build their own su­pe­rior ‘chil­dren’. Unchecked by the gla­cial mores of evo­lu­tion or the lim­i­ta­tions of mere bi­o­log­i­cal mat­ter, it is un­sur­pris­ing that sci- prophecy fore­sees mankind very soon at the mercy of his meis­ter­w­erk.

The tip­ping point – the point that Tran­scen­dence imag­ines – is known to com­puter sci­en­tists as the sin­gu­lar­ity. It is the sin­gle mo­ment when AI over­takes hu­man in­tel­li­gence. That is to say com­put­ers will not only be able to beat us at chess, but they will be able to process a suit­ably vic­to­ri­ous emo­tional re­sponse, too. Af­ter all, our own brains are sim­ply a bi­o­log­i­cal switch­ing sys­tem with synapses for tran­sis­tors. Why should a com­puter not feel re­ward too?

The sin­gu­lar­ity the­ory as­sumes we can reach – and go be­yond – such a point, and that isn’t pie-in-the-sky idea born in the writ­ers’ room at Warner Bros. It is based on es­tab­lished sci­enti c re­search, and it could come along sooner than you might think.

Ray Kurzweil is an in­ven­tor, au­thor and fu­tur­ist. He’s also Google’s di­rec­tor of en­gi­neer­ing. Oh, and he thinks the sin­gu­lar­ity could hap­pen in the 2040s, thirty-odd years from now. He sees a fu­ture in which com­put­ers are a bil­lion times more pow­er­ful than the hu­man brain. In case you were in any doubt about whether he would stick by those claims, he’s even writ­ten them in a book called The Sin­gu­lar­ity Is Near. If his sta­tion at Google isn’t enough to con­vince you of his san­ity, he was granted Amer­ica’s big­gest tech medal by Bill Clin­ton in 1999. See the pho­to­copier in your of ce? Yeah, he in­vented the atbed scan­ner, too.

That’s not to say Kurzweil isn’t out­spo­ken. But he’s also not alone. And this is where the ctional plot of Tran­scen­dence thick­ens into re­al­ity. Last year in New York sci­en­tists gath­ered for the Global Fu­ture 2045 con­fer­ence, a net­work­ing event de­signed to bring to­gether the minds that hope to bring about hu­man im­mor­tal­ity by com­bin­ing our bi­o­log­i­cal brains with the in­fal­li­bil­ity of mod­ern com­put­ing.

The con­fer­ence was founded by Dmitry It­skov, a Rus­sian en­tre­pre­neur who is re­puted to have spent US 3 mil­lion on the project to date. The plan is to

The com­put­ers that drive cars and beat you at chess will one day be able to learn bet­ter and faster than you

de­velop first a ro­bot that can be con­trolled via a hu­man brain, then a method of trans­fer­ring a bi­o­log­i­cal brain into such a ro­bot. Once these mile­stones have been reached, it will be time to work on trans­fer­ring data from an or­ganic brain to a syn­thetic one. Lastly, the sci­en­tists will at­tempt to cre­ate holo­graphic be­ings to re­place cum­ber­some phys­i­cal ro­bots.

Mad as all this sounds, some of the tech­nolo­gies al­ready ex­ist. Pros­thet­ics can al­ready be con­trolled us­ing nerve trig­gers – down to the move­ment of a sin­gle nger. Keep­ing the brain alive by ar­ti­fi­cial means is an es­tab­lished med­i­cal prac­tice. And Tu­pac Shakur rose from the dead to per­form as a holo­gram along­side Snoop Dog and Dr Dre at Coachella, a Californian mu­sic and arts fes­ti­val, in 2012. So while the idea of holo­grams wan­der­ing about with syn­thetic hu­man brains (mem­o­ries and all) might seem like some­thing that be­longs in the next century, It­skov is aim­ing to get to that stage by 2045, just over 30 years from now.

At the Global Fu­ture 2045 con­fer­ence, roboti­cists can rub shoul­ders with heavy­weight hu­man­i­tar­ian foun­da­tion lead­ers and the likes of Ray Kurzweil (he at­tended last

year’s event). From afar, the Dalai Lama has en­dorsed It­skov’s en­deav­our, while closer to home the US govern­ment is in­vest­ing US$100 mil­lion in sci­ence to “bet­ter un­der­stand how we think, learn and re­mem­ber” and back­ing the De­fense Ad­vanced Re­search Projects Agency (DARPA) with a fur­ther US$50 mil­lion for “un­der­stand­ing the dy­namic func­tions of the brain.” DARPA has al­ready de­vel­oped a hu­manoid ro­bot called At­las, which is ex­pected to be able to drive a car and op­er­ate power tools this year.

Are we all sit­ting com­fort­ably now?

More con­cern­ing still are the nu­mer­ous unan­swered eth­i­cal ques­tions, some of which are ad­dressed in Tran­scen­dence. Does a hu­man brain that has been trans­ferred to a ma­chine qual­ify for the same rights as any other per­son? What if it is sim­ply the data that has been trans­ferred – mem­o­ries, mo­tor skills, knowl­edge? How can we pro­tect our species against a cre­ation more in­tel­li­gent than us? What hap­pens, in short, if things start to go wrong? Or should we say when things go wrong?

Un­sur­pris­ingly, Tran­scen­dence as­sumes a cau­tion­ary po­si­tion on the pos­si­bil­i­ties that await us on the other side of the sin­gu­lar­ity, as do the bulk of sci- sto­ries in its vein (yes, even Franken­stein). But that’s just be­cause it makes for a bet­ter plot. Right?

Well, yes. And no. There is plenty of good-na­tured ar­ti­fi­cial in­tel­li­gence in the sci- genre. In Dou­glas Adams’ Hitch­hik­ers Guide To The Galaxy, a com­puter called Deep Thought builds a su­pe­rior com­puter called Earth. It is so large that it is of­ten mis­taken for a planet and is fa­mously de­scribed in the epony­mous in­ter­plan­e­tary guide as ‘mostly harm­less.’ The Cul­ture uni­verse, a sci- en­vi­ron­ment cre­ated by Iain M Banks, is a post-hu­man lib­eral an­ar­chy where ar­ti­fi­cial in­tel­li­gence be­nignly pro­vides a life­style of su­per­nat­u­ral abun­dance for its sen­tient sub­jects.

But these are the rare ex­cep­tions to the usual as­sump­tion that once we have cre­ated su­pe­rior ro­bots, they will turn on us. Sci- god­fa­ther Isaac Asi­mov fa­mously pos­tu­lated the Three Laws of Ro­bot­ics to be pro­grammed into the DNA of ad­vanced AI to pre­vent ro­bots hurt­ing hu­mans or each other, and to en­sure they would obey our in­struc­tions. But he also man­aged to en­vis­age a sit­u­a­tion in which a ro­bot with slightly mod­i­fied cod­ing could jus­tify at­tack­ing a hu­man. (The short story, en­ti­tled Lit­tle Lost Ro­bot, formed the ba­sis for the 2004 lm I, Ro­bot.)

An­other com­mon sce­nario in which we find our­selves at war with our ar­ti­fi­cially in­tel­li­gent cre­ations sees us

Does a hu­man brain that has been trans­ferred to a ma­chine qual­ify for the same rights as any other per­son?

com­pet­ing for re­sources in a stark re­ver­sal of the fu­ture en­vis­aged in Ian M Banks’ Cul­ture uni­verse. But that as­sumes that tech­nol­ogy will be com­pet­i­tive, a trait that is un­likely to cross the ar­ti­fi­cial mind of some­thing that has not had to en­dure the gru­elling chal­lenges of ge­netic evo­lu­tion.

More re­al­is­ti­cally, ar­ti­fi­cial in­tel­li­gence could be­come the next nu­clear war­fare with gov­ern­ments or in­di­vid­u­als (pre­sum­ably the sort of evil overlords de­picted by Bond lms) mis­ap­pro­pri­at­ing the tech­nol­ogy for their own per­sonal gain. It’s pretty har­row­ing to think that the world’s most feared weapon of mass de­struc­tion might be able to think for it­self, even build more of it­self.

If all this seems too ter­ri­fy­ing to imag­ine, re­lax. We have a ma­jor de­fence against our real-life Franken­steins, should they turn against us in our life­times. It is the uni­ver­sal saviour of tech­no­log­i­cal re­bel­lion, as any IT helpdesk can al­ready tell you. This safety mea­sure has been in­stalled on al­most ev­ery elec­tronic de­vice ever made and continues to be an im­por­tant phys­i­cal fea­ture of gad­getry, even in our touch­screen-ob­sessed world. We call it the off switch.

English ac­tor Peter Cush­ing as Baron Vic­tor Franken­stein in ‘ Franken­stein Must Be De­stroyed’, 1969

Ray Kurzweil

Colin Cliv plays the driven doc­tor and Dwight Frye plays his de­formed as­sis­tant Fritz in Franken­stein, 1931, di­rected by James Whale Johnny Depp in Tran­scen­dence

A Space Odyssey dur­ing film­ing with di­rec­tor Stan­ley Kubrick lin­ing up shot through cam­era Yaphet Kotto, Sigour­ney Weaver and Ian Holm on the set of Ri­d­ley Scott’s sci­ence fic­tion clas­sic Alien, 1979

Dun­can Jones’ Moon, 2009

Newspapers in English

Newspapers from UAE

© PressReader. All rights reserved.