Meet­ing of mil­i­tary minds the first shot in AI arms race

The US gath­er­ing could pro­vide a foun­da­tion for the West to com­pete with China on ma­chine learn­ing

The Daily Telegraph - Business - - Business Comment - Garry white

In the fu­ture, ma­jor de­ci­sions taken in the the­atre of war are likely to be made by ma­chines. The de­ci­sion to shoot a mis­sile or drop a bomb will be made elec­tron­i­cally, elim­i­nat­ing the flawed hu­man mind from any de­ci­sion-mak­ing process. Sys­tems based on ar­ti­fi­cial in­tel­li­gence will per­form many of the tasks that are cur­rently the re­spon­si­bil­ity of hu­mans.

These ro­bot “brains” will gather and as­sess real-time data, scru­ti­n­is­ing facts to de­cide the next course of action. This could ul­ti­mately lead to full weapons sys­tems – even nu­clear war­heads – be­ing con­trolled by an un­emo­tional, in­de­pen­dent box of elec­tronic tricks. Clearly, the eth­i­cal con­sid­er­a­tions are im­mense.

This week, of­fi­cials from 13 US-al­lied coun­tries met on­line to try to fig­ure out how they could use AI and ma­chine learn­ing across their mil­i­tary and de­fence ca­pa­bil­i­ties, but in such a way as to blunt the nu­mer­ous po­ten­tial down­sides.

The meet­ing was or­gan­ised by the US Joint Ar­ti­fi­cial In­tel­li­gence Cen­tre (JAIC) on be­half of the de­fence de­part­ment’s AI Cen­tre of Ex­cel­lence. It aims to find a way to in­te­grate rapid paces of tech­no­log­i­cal change with a solid foun­da­tion of eth­i­cal rules and reg­u­la­tions. It will also share knowl­edge and de­velop uni­fied pro­cesses and data be­tween its mem­bers. Strong re­la­tion­ships will be es­sen­tial to en­sure its long-term suc­cess and Amer­ica’s al­lies will un­der­stand this too.

Not only is this ini­tia­tive im­por­tant for the safety and se­cu­rity of fu­ture gen­er­a­tions, it is also po­lit­i­cally sig­nif­i­cant to­day. Washington has clashed with many al­lies dur­ing the Trump pres­i­dency on nu­mer­ous geopo­lit­i­cal is­sues – even rais­ing trade bar­ri­ers on Canada, Mex­ico, the Euro­pean Union and Ja­pan. But man­ag­ing the fu­ture of AI in war­fare is an easy ban­ner on which to unite.

AI is con­cerned with the devel­op­ment of smart ma­chines that can per­form the com­plex tasks typ­i­cally as­so­ci­ated with hu­man in­tel­li­gence. AI can be used for in­for­ma­tion gath­er­ing, sur­veil­lance and re­con­nais­sance – but could also be con­nected to live weapons sys­tems. If ma­chines are handed the abil­ity to make de­ci­sions that could re­sult in death and de­struc­tion, the global se­cu­rity im­pli­ca­tions are enor­mous.

This week’s sum­mit was the first to try to man­age this tech­no­log­i­cal progress from an eth­i­cal point of view. Par­tic­i­pat­ing coun­tries in­cluded mil­i­tary del­e­ga­tions from Aus­tralia, Canada, Den­mark, Es­to­nia, Fin­land, France, Is­rael, Ja­pan, Nor­way, the Repub­lic of Korea, Swe­den and the UK.

“We want it to be al­most like a prob­lem-solv­ing fo­rum,” Stephanie Cul­ber­son, di­rec­tor of in­ter­na­tional AI pol­icy at the JAIC, noted. She said that the meet­ings were de­signed to be in­for­mal and col­lab­o­ra­tive – un­like the usual mil­i­tary-to-mil­i­tary en­gage­ments, which are highly for­mal. The fo­rum will also act as a con­duit for shar­ing tech­ni­cal in­for­ma­tion – with con­sis­tent stan­dards ap­plied be­tween these al­lies on the man­age­ment of data – the lifeblood of ma­chine learn­ing and AI.

China was ob­vi­ously not present at the eth­i­cal dis­cus­sion. AI is an­other fron­tier field where progress is be­ing ac­cel­er­ated by the ide­o­log­i­cal clash be­tween Washington and Bei­jing. Each coun­try wants to at­tain global lead­er­ship in the tech­nol­ogy that will drive the economies of to­mor­row and the AI arms race is a ma­jor bat­tle front.

China’s rapid suc­cess in this area is con­cern­ing. Bei­jing’s au­thor­i­tar­ian gov­ern­ment uses AI and cit­i­zens’ data in ways that are im­pos­si­ble in demo­cratic coun­tries be­cause they would vi­o­late pri­vacy and civil lib­erty laws. Fa­cial recog­ni­tion tech­nolo­gies used for the sur­veil­lance and de­ten­tion of Mus­lim eth­nic mi­nori­ties in its west­ern Xin­jiang prov­ince have also been a drive of its in­no­va­tion in this area. Courts in the coun­try have even started us­ing AI-gen­er­ated

as­sess­ments to help gen­er­ate sen­tenc­ing de­ci­sions.

China de­clared in 2017 that it wanted to be the world leader in AI by 2030. While the US still leads in ab­so­lute terms, China has thrown tens of bil­lions of dol­lars on its AI pro­grammes and is di­rect­ing com­pa­nies in the spe­cific ar­eas that the cen­tral gov­ern­ment deems a pri­or­ity. Amer­ica’s “mod­est” lead over China in AI re­mains, but that is mainly be­cause of its ad­vanced semi­con­duc­tor sec­tor, ac­cord­ing to a study by US think tank the Rand Cor­po­ra­tion. US lead­er­ship in AI is there­fore de­pen­dent on its lead­er­ship in semi­con­duc­tor wafers – and China re­cently un­veiled its plan to leapfrog Sil­i­con Val­ley in this area too. China will specif­i­cally write this aim into its 14th five-year plan.

The Rand re­port warned that Bei­jing’s use of cen­tral plan­ning has given it an ad­van­tage in its at­tempt to beat Sil­i­con Val­ley through its abil­ity to pull the strings on cor­po­rate busi­ness de­ci­sions. A ma­jor ad­van­tage is also found in the coun­try’s vast pop­u­la­tion. China’s pop­u­la­tion is about four times that of the US – and AI de­vel­op­ers are more likely to get

‘AI can be used for in­for­ma­tion gath­er­ing and sur­veil­lance – but could also be con­nected to live weapons sys­tems’

their hands on vast and com­plete sets of data that drive AI devel­op­ment. In the pri­vacy-lov­ing West, re­searchers are un­likely to be given such a use­ful devel­op­ment tool.

“We should lever­age our so­cial­ist sys­tem’s ad­van­tage in con­cen­trat­ing re­sources to get ma­jor un­der­tak­ings done, and achieve break­throughs in key and core tech­nolo­gies,” China’s pres­i­dent Xi Jin­ping said last month.

The JAIC meet­ing held this week could pro­vide a good foun­da­tion for the West to com­pete with China on AI, if all goes to plan. Pool­ing clean data sets be­tween coun­tries will give re­searchers in­for­ma­tion at a bet­ter scale – and global in­for­ma­tion gath­er­ing can be stan­dard­ised to man­age the data ef­fi­ciency and pro­vide it in use­ful, clean for­mats. It is also a fo­rum to dis­cuss the dan­ger­ous eth­i­cal is­sues that will cer­tainly emerge. But, fol­low­ing a few tur­bu­lent years be­tween friends, the fact Amer­ica’s al­lies can eas­ily rally to Washington’s call on this is­sue is pos­i­tive for the free world.

Garry White is chief in­vest­ment com­men­ta­tor at wealth man­age­ment com­pany Charles Stan­ley

The next step from hu­man-con­trolled ro­bots is for au­ton­o­mous weapons and tools to make their own de­ci­sions

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.