The fu­ture is com­ing but not as we think it

The Guardian Weekly - - Books - Life 3.0 by Max Teg­mark Allen Lane, 384pp Yuval Noah Harari

If you hear a sce­nario about the world in 2050 and it sounds like sci­ence fic­tion, it is prob­a­bly wrong; but if you hear a sce­nario about the world in 2050 and it does not sound like sci­ence fic­tion, it is cer­tainly wrong.

Tech­nol­ogy is never de­ter­min­is­tic: it can be used to cre­ate very dif­fer­ent kinds of so­ci­ety. In the 20th cen­tury, trains, elec­tric­ity and ra­dio were used to fash­ion Nazi and com­mu­nist dic­ta­tor­ships, but also to foster lib­eral democ­ra­cies and free mar­kets. In the 21st cen­tury, AI will open up an even wider spec­trum of pos­si­bil­i­ties. De­cid­ing which of th­ese to re­alise may well be the most im­por­tant choice hu­mankind will have to make in the com­ing decades. This choice is not a mat­ter of en­gi­neer­ing or sci­ence. It is a mat­ter of pol­i­tics. Hence it is not some­thing we can leave to Sil­i­con Val­ley – it should be among the most im­por­tant items on our po­lit­i­cal agenda. Un­for­tu­nately, AI has so far hardly reg­is­tered on our po­lit­i­cal radar.

Max Teg­mark’s Life 3.0 tries to rec­tify the sit­u­a­tion. Writ­ten in an ac­ces­si­ble and en­gag­ing style, the book of­fers a po­lit­i­cal and philo­soph­i­cal map of the prom­ises and per­ils of the AI revo­lu­tion. In­stead of push­ing any one agenda or pre­dic­tion, Teg­mark seeks to cover as much ground as pos­si­ble, re­view­ing a wide va­ri­ety of sce­nar­ios con­cern­ing the im­pact of AI on the job mar­ket, war­fare and po­lit­i­cal sys­tems.

Life 3.0 does a good job of clar­i­fy­ing ba­sic terms and key de­bates, and in dis­pelling com­mon myths. While sci­ence fic­tion has caused many peo­ple to worry about evil ro­bots, for in­stance, Teg­mark rightly em­pha­sises that the real prob­lem is with the un­fore­seen con­se­quences of de­vel­op­ing highly com­pe­tent AI. In Teg­mark’s words, “the real risk with ar­ti­fi­cial gen­eral in­tel­li­gence isn’t mal­ice but com­pe­tence. A su­per­in­tel­li­gent AI will be ex­tremely good at ac­com­plish­ing its goals, and if those goals aren’t aligned with ours, we’re in trou­ble.”

Nat­u­rally Teg­mark’s map is not com­plete, and in par­tic­u­lar it does not give enough at­ten­tion to the con­flu­ence of AI with biotech­nol­ogy. The 21st cen­tury will be shaped not by in­fotech alone, but rather by the merger of in­fotech with biotech. AI will be of cru­cial im­por­tance pre­cisely be­cause it will give us the com­put­ing power nec­es­sary to hack the hu­man or­gan­ism. Long be­fore the ap­pear­ance of su­per­in­tel­li­gent com­put­ers, our so­ci­ety will be com­pletely trans­formed by rather crude and dumb AI that is nev­er­the­less good enough to hack hu­mans, pre­dict their feel­ings, make choices on their be­half and ma­nip­u­late their de­sires. It might be apoc­a­lypse by shop­ping.

Yet the real prob­lem of Teg­mark’s book is that it soon bumps up against the lim­its of present-day po­lit­i­cal de­bates. The AI revo­lu­tion turns many philo­soph­i­cal prob­lems into prac­ti­cal po­lit­i­cal ques­tions and forces us to en­gage in “phi­los­o­phy with a dead­line” (as the philoso­pher Nick Bostrom called it). Philoso­phers are pa­tient peo­ple, en­gi­neers are im­pa­tient, and hedge fund in­vestors are more rest­less still. When Tesla en­gi­neers come to de­sign a self-driv­ing car, they can­not wait while philoso­phers ar­gue about its ethics.

Con­se­quently Teg­mark soon leaves be­hind fa­mil­iar de­bates about jobs, pri­vacy and weapons of mass de­struc­tion, and ven­tures into realms that hith­erto were as­so­ci­ated with phi­los­o­phy, the­ol­ogy and mythol­ogy, tak­ing things be­yond our own planet. This can hardly be avoided but I fear that many of his prospec­tive read­ers will not fol­low him there. Our po­lit­i­cal sys­tems, and in­deed our in­di­vid­ual minds, are just not built to think on such a scale.

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.