Re­al­iz­ing AI po­ten­tial will take time

Global Times - - Business -

Edi­tor’s Note:

There’s no hot­ter phrase than ar­ti­fi­cial in­tel­li­gence (AI) in today’s tech­nol­ogy world. This is es­pe­cially so in China, where Google’s Go-play­ing AI Al­phaGo trounced top-ranked Chi­nese Go grand­mas­ters at a Go sum­mit in Wuzhen, East China’s Zhe­jiang Prov­ince in May.

Adding fuel to China’s AI hype, Bei­jing-based AI chip start-up Cam­bri­con Tech­nolo­gies set a tar­get of hav­ing 1 bil­lion de­vices glob­ally us­ing its chips while un­veil­ing new AI pro­ces­sors ear­lier in Novem­ber.

In an exclusive in­ter­view in Bei­jing with Global Times re­porter Li Qiaoyi (GT) last week, Lin Chenxi (Lin), co-founder of Shang­hai-based YITU Tech, one of China’s ma­jor AI start-ups, re­vealed that the com­pany has in­vested in a do­mes­tic chip­maker to en­able in­creases in AI com­put­ing power. The for­mer se­nior ex­pert at Aliyun Com­put­ing, Alibaba’s cloud-com­put­ing arm, also shared his thoughts on the prospects for the coun­try’s AI space.

GT: What ar­eas and in­dus­tries do you think have great po­ten­tial in terms of the prac­ti­cal ap­pli­ca­tions of AI tech­nolo­gies but have yet to be widely rec­og­nized as a fer­tile field for AI ap­pli­ca­tions?

Lin: Even where there al­ready seems to be a lot of hype about AI ap­pli­ca­tions, there’s still a long way to go be­fore AI plays a sig­nif­i­cant role in re­ju­ve­nat­ing any given in­dus­try. For ex­am­ple, along­side the grow­ing aware­ness of AI’s ca­pa­bil­i­ties in the pub­lic safety and se­cu­rity arena, the health­care sec­tor has over the past year been in­creas­ingly considered a race track for busi­nesses mak­ing bets on AI.

Nev­er­the­less, there is still much un­seen po­ten­tial for AI’s ap­pli­ca­tions in the health­care sec­tor. That will mean an im­prove­ment not only in the ef­fi­ciency of treat­ing ill­nesses but also the ef­fi­ciency of doc­tors learn­ing about how dis­eases can be cured, as ma­chines are seen do­ing a splen­did job in gath­er­ing hu­man physi­cians’ ex­pe­ri­ence and mak­ing them­selves into ex­perts based on ma­chine learn­ing.

This will re­quire tech­nol­ogy break­throughs not just in com­puter vi­sion, but in nat­u­ral lan­guage pro­cess­ing, se­man­tic com­pre­hen­sion and health­care knowl­edge map­ping. Data ac­cu­mu­lated over the course of un­leash­ing the po­ten­tial in AI’s health­care ap­pli­ca­tions could also in­flu­ence drugs’ de­vel­op­ment so as to ad­dress even more prob­lems.

Down the road, the new re­tail, ed­u­ca­tion and trans­port in­dus­tries, as well as man­u­fac­tur­ing and other ar­eas re­lated to peo­ple’s liveli­hoods are ex­pected to be re­made by AI.

GT: Are China’s AI com­pa­nies over­tak­ing their US coun­ter­parts in terms of tech­no­log­i­cal strength and over­all com­pet­i­tive­ness?

Lin: To be sure, China’s AI start-ups at large are barely on the same level as the likes of Google, al­though some Chi­nese AI com­pa­nies might prove to be stronger than Google in cer­tain tech­nolo­gies such as face recog­ni­tion. Even Chi­nese tech­nol­ogy com­pa­nies as big as Baidu, Alibaba and Ten­cent are still not on par with Google as mea­sured by AI vi­sion, tal­ent pools and over­all tech­no­log­i­cal strength.

That said, with strong sup­port from the Chi­nese gov­ern­ment, which con­sid­ers AI to be pow­er­ing the coun­try’s fu­ture growth, China has made huge head­way in the AI space. It is be­lieved that China and the US will be­come the twin en­gines pow­er­ing the global AI sec­tor in the next decade.

Also worth point­ing out is that while China has ap­par­ently taken bolder moves in cre­at­ing in­no­va­tive AI ap­pli­ca­tions, it is wrong to think China has in­no­vated only in ap­pli­ca­tion terms. For in­stance, there is an ac­tual de­mand in China for fa­cial recog­ni­tion tech­nol­ogy to be widely used, con­sid­er­ing the need to spot crim­i­nal sus­pects among our huge pop­u­la­tion, among other sce­nar­ios. Prac­ti­cal needs will prob­a­bly re­sult in the­o­ret­i­cal break­throughs, which will help in ad­dress­ing world­class prob­lems.

Even­tu­ally, there can be break­throughs in al­go­rithms and soft­ware ar­chi­tec­ture. Global prizes, es­pe­cially in fa­cial recog­ni­tion tech­nolo­gies that have been won by do­mes­tic AI com­pa­nies in­clud­ing YITU Tech, sig­nal the coun­try’s ris­ing prow­ess in the AI world.

GT: Can you com­pare the AI start-up en­vi­ron­ments in China and the US?

Lin: China drew up an AI de­vel­op­ment plan ear­lier this year that sees the coun­try as be­ing a world leader in AI by 2030. AI de­vel­op­ment, as such, has be­come a na­tional plan, which means mas­sive gov­ern­ment fund­ing sup­port, more in­cen­tives for ci­ties to at­tract AI tal­ent and greater me­dia cov­er­age of AI-re­lated ini­tia­tives.

This is quite dif­fer­ent from a few years ago when China’s tech­nol­ogy start-ups pri­mar­ily fo­cused on the mo­bile In­ter­net, par­tic­u­larly the on­line-to-off­line busi­nesses. At that time, a Ts­inghua Univer­sity grad­u­ate would have hes­i­tated to join an AI com­pany, as the con­cept was hardly known. But it’s en­tirely dif­fer­ent now as the gov­ern­ment’s AI push has served to sub­stan­tially el­e­vate pub­lic recog­ni­tion of AI-re­lated po­si­tions. This makes a lot of sense not just in re­cruit­ing AI tal­ent, but in mar­ket­ing and reach­ing out to the mass mar­ket. In this re­gard, China has built a bet­ter en­vi­ron­ment than Sil­i­con Val­ley, with do­mes­tic start-ups hav­ing more ac­cess to Se­ries A round fund­ing, which en­ables a more vi­brant AI start-up en­vi­ron­ment in China.

GT: How do you eval­u­ate do­mes­tic AI start-ups’ abil­ity to mon­e­tize the tech­nol­ogy?

Lin: Un­like many mar­ket ob­servers who doubt AI busi­nesses’ abil­ity to make money, I be­lieve it’s ac­tu­ally not that hard for an AI busi­ness to do so, par­tic­u­larly in the busi­ness-to-busi­ness arena. In such cases, AI ap­pli­ca­tions in spe­cific sce­nar­ios make it easier for mon­e­ti­za­tion to be achieved once cer­tain tech­no­log­i­cal break­throughs re­shape an in­dus­try.

For ex­am­ple, AI-pow­ered con­sumer ser­vice sys­tems can be cre­ated to re­make the tra­di­tion­ally hu­man-in­ten­sive con­sumer ser­vice busi­nesses. It could be slow at first, but it will grow very quickly at a com­pound rate. In con­sumer-to-con­sumer ar­eas, it will take more time for mon­e­ti­za­tion to be achieved, as gen­uine “killer apps” ap­ply­ing AI tech­nolo­gies have yet to be seen.

GT: How far will AI even­tu­ally go? What’s YITU Tech’s AI vi­sion?

Lin: It’s im­pos­si­ble to fore­see what AI would achieve in tech­no­log­i­cal terms. Take Al­phaGo as an ex­am­ple. At a cer­tain point, ma­chines will level the play­ing field with hu­mans, but be­yond that point, ma­chines will beat hu­mans by a large mar­gin. I wouldn’t say tech­no­log­i­cal ad­vance­ments will one day de­stroy hu­mans, but it still de­pends in whose hands the tech­nolo­gies are. It’s ac­tu­ally the same with nu­clear weapons. There need to be rules that AI can’t be used to do evil things.

In the case of YITU, in ad­di­tion to AI chip in­vest­ment, it will be of grow­ing im­por­tance for the com­pany to ex­plore over­seas mar­kets. The com­pany has just signed a strate­gic agree­ment with the Sin­ga­pore gov­ern­ment about pro­vid­ing AI-pow­ered safety and se­cu­rity so­lu­tions. Look­ing for­ward, Euro­pean coun­tries, African coun­tries and mar­kets along the route of the Belt and Road ini­tia­tive will be key ar­eas to be tapped.

Photo: Courtesy of YITU Tech

Lin Chenxi

Newspapers in English

Newspapers from China

© PressReader. All rights reserved.