There are cer­tainly dooms­day sce­nar­ios that could arise if we cross the bridge to set­ting AI loose in the world.

ArabAd - - COVER STORY -

There are cer­tainly dooms­day sce­nar­ios that could arise if we cross the bridge to set­ting AI loose in the world, but as long as AI sys­tems are over­seen and con­trolled by hu­mans we should be fine

AI will need to be de­vel­oped re­spon­si­bly within and along­side hu­man so­ci­ety and liability ap­pro­pri­ately al­lo­cated be­tween its de­sign­ers, its op­er­a­tors, and (per­haps even­tu­ally) the AI it­self.

When we look at the mar­ket­ing uni­verse to­day, the role of the CMO has al­ready been rev­o­lu­tionised by AI. There have never been more chan­nels with which a mar­keter can reach their tar­get mar­ket, nor have CMOS had so much data avail­able to them with which they can use to make de­ci­sions. There is too much com­pli­cated data for hu­mans to man­age. Com­put­ers are able to an­a­lyse the data, recog­nise pat­terns, learn from all of this in­for­ma­tion and act upon what they have learnt – quickly. Within mil­lisec­onds, a com­puter can recog­nise a be­havioural pat­tern and re­act, in or­der to cap­i­talise on op­por­tu­ni­ties as they hap­pen. The pro­pri­etary al­go­rithm from Rocket Fuel can eval­u­ate nearly a mil­lion ad impressions ev­ery sec­ond and make 1.28 bil­lion de­ci­sions per day.

Be­cause of th­ese pos­si­bil­i­ties many are ap­pre­hen­sive to make the shift to ar­ti­fi­cial in­tel­li­gence. How do you com­bat fear of ai that might stem from th­ese po­ten­tial cir­cum­stances?

A pes­simistic view of bur­geon­ing new tech­nolo­gies is noth­ing new. The his­tory of de­vel­op­ment of communications and com­put­ing tech­nol­ogy has shown that each new de­par­ture is met with fear and crit­i­cism.

Cur­rently the main im­mi­nent threat from AI is the po­ten­tial that many of the jobs that re­quire hu­man at­ten­tion and de­ci­sion mak­ing to­day may be per­formed bet­ter, faster, cheaper in the fu­ture by ro­bots or AI soft­ware sys­tems. It is a chal­lenge for us to find a way to ad­just the econ­omy in light of this new low-cost pres­sure on work­ers and wages.

But if you think about the types of jobs that peo­ple have th­ese days, a large num­ber of them didn’t ex­ist a decade or two ago. As jobs be­come au­to­mated, there will con­cur­rently be new jobs cre­ated to cope with the new way so­ci­ety is op­er­at­ing. Peo­ple will be freed from those ‘back-break­ing’ and un­de­sir­able jobs and in­stead will be able to un­der­take roles that al­low them to reach their po­ten­tial.

Most im­por­tantly AI will free up hu­man cre­ative po­ten­tial and cre­ate a myr­iad of new roles and op­por­tu­ni­ties that we can’t even be­gin to imag­ine.

We of­ten see ex­am­ples of in­ef­fi­cient tar­get­ing such as ads for ho­tel rates dur­ing or af­ter a trip or ads for a lap­top af­ter a pur­chase among oth­ers. This is hardly ef­fi­cient use of ad space. Do buy­ers voice their con­cerns on this topic and on which side of the chain should this and other ex­am­ples of in­ef­fi­cient tar­get­ing be ad­dressed?

Yes, in­ef­fi­cient tar­get­ing when it hap­pens is not only an­noy­ing for con­sumers, it also wastes mar­ket­ing bud­gets and un­der­mines the brand’s rep­u­ta­tion.

But many tech providers like Rocket Fuel are al­ready able to very pre­cisely tar­get users in the right mo­ment of in­flu­ence with the right ad.

Site per­for­mance data can give in­sights into anony­mous on­line be­havioural pat­terns of past shop­pers, Rocket Fuel’s Mo­ment Scor­ing tech­nol­ogy then analy­ses con­sumers' be­hav­iour in or­der to pre­dict a buy­ing in­ten­tion and de­ter­mine which ad­vert should be dis­played and use a dy­namic cre­ative to dis­play the most ef­fec­tive mes­sage.

For ex­am­ple, when an ad­ver­tis­ing cam­paign for a city ho­tel des­ti­na­tion is launched, the al­go­rithm anony­mously analy­ses the in­ter­net be­hav­iour of ex­ist­ing cus­tomers who have al­ready pur­chased ac­com­mo­da­tion, via pix­els on the ad­ver­tiser’s web­site. The al­go­rithm is then able to iden­tify con­sumers who have shown the same on­line be­hav­iour in re­cent weeks as th­ese ex­ist­ing cus­tomers. The al­go­rithm pre­dicts whether a con­sumer is in­ter­ested in book­ing a ho­tel room and will only serve ads to con­sumers fit­ting this pro­file, sig­nif­i­cantly in­creas­ing the rel­e­vancy and ef­fec­tive­ness of this ad­ver­tis­ing and the re­turn on in­vest­ment for this cam­paign. Ex­ist­ing cus­tomers can be tar­geted with com­pli­men­tary of­fers, like tours or restau­rant book­ings, based on their on­line be­havioural pat­terns and in­ter­ests.

Newspapers in English

Newspapers from Bahrain

© PressReader. All rights reserved.