‘AI back­lash likely if it’s not prop­erly ex­plained’

Western Mail - - NEWS -

GOV­ERN­MENTS and tech com­pa­nies risk a pop­u­lar back­lash against ar­ti­fi­cial in­tel­li­gence (AI) un­less they bet­ter ex­plain how it will be used, ac­cord­ing to a new re­port.

Polling for the Royal So­ci­ety of Arts (RSA) found wide­spread con­cern that AI will cre­ate a “Com­puter Says No” cul­ture, in which cru­cial de­ci­sions are made au­to­mat­i­cally with­out con­sid­er­a­tion of in­di­vid­ual cir­cum­stances.

If the pub­lic feel “vic­timised or dis­em­pow­ered” by in­tel­li­gent ma­chines, they may re­sist the in­tro­duc­tion of new tech­nolo­gies, even if it holds back progress which could ben­e­fit them, the re­port warned.

Fear of in­flex­i­ble au­to­matic de­ci­sion-mak­ing was a greater con­cern than ro­bots tak­ing hu­mans’ jobs among those tak­ing part in a sur­vey by poll­sters YouGov for the RSA.

De­spite re­cent pub­lic­ity about the mis­use of Face­book users’ per­sonal data to tar­get on­line ads, is­sues sur­round­ing so­cial mar­ket­ing were bot­tom of the list of con­cerns.

Some 60% of those ques­tioned op­posed au­to­mated de­ci­sion-mak­ing by com­put­ers in re­cruit­ment and pro­mo­tion choices, and the same pro­por­tion said it should not be used to help courts judge whether to grant a de­fen­dant bail or rec­om­mend re­ha­bil­i­ta­tion. Just 11% backed its use in re­cruit­ment and 12% in the courts.

Use of au­to­mated de­ci­sion-mak­ing in the im­mi­gra­tion sys­tem was op­posed by a mar­gin of 54%-16%, in so­cial se­cu­rity by 52%-17%, in health­care by 48%-20% and in fi­nan­cial ser­vices by 48%-27%.

Only in the area of ad­ver­tis­ing and so­cial me­dia were opin­ions more bal­anced, with 28% op­pos­ing its use against 26% who sup­ported it. Some 61% said their main con­cern about au­to­mated de­ci­sion-mak­ing was AI does not have “the com­pas­sion to make im­por­tant de­ci­sions”.

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.