0
Interview
1 December 2019 2.00 h

Begoña Román

Chair of the Ethical Committee of Social Services in Catalonia

“AI is made so that you like it”

Does the dif­fer­ence be­tween types of in­tel­li­gence lie in emo­tion?
Ex­actly. Peo­ple think about ro­bots. Yet, most AI has noth­ing to do with ro­bots. It’s ar­ti­fi­cial in­tel­li­gence we use to del­e­gate tasks that hu­mans used to do and now leave in the hands of al­go­rithms de­signed by hu­mans. They are in­creas­ingly so­phis­ti­cated al­go­rithms that in­clude the pro­gram­mer’s val­ues. There­fore, the ethics of ar­ti­fi­cial in­tel­li­gence are those of the pro­gram­mer.
What are the eth­i­cal im­pli­ca­tions?
In the car in­dus­try, for ex­am­ple, if a brand had to de­cide whose life to pri­ori­tise, it might choose to put the dri­ver’s safety first be­cause that is who buys the car. That would be de­cided by a pro­gram­mer, based on the de­ci­sion made by the com­pany. An­other might choose to de­sign sen­sors that can de­tect if there’s a child in the car, and so pro­tect it even if it might mean the death of the dri­ver. The al­go­rithm in­cludes the pro­gram­mer’s and the com­pany’s val­ues. Al­go­rithms can deal with a lot of data that the human brain — not due to the com­plex­ity but to the quan­tity of data — can­not process quickly enough. That means you can pro­vide the com­puter with cri­te­ria to make de­ci­sions that re­quire huge amounts of data. An ex­am­ple. They came up with some pro­grams to help ra­di­ol­o­gists de­tect ill­nesses with scan­ners. Both hu­mans and ma­chines showed a 5% mar­gin of error, but to­gether the mar­gin dropped to 1%.
Is the fu­ture about find­ing the bal­ance be­tween the two in­tel­li­gences?
AI is a prod­uct of human in­tel­li­gence and it would be a shame to let it get out of hand, es­pe­cially con­sid­er­ing the risk we are ex­posed to when there is no col­lab­o­ra­tion. Al­go­rithms are every­where, which is why you are often asked on­line to show you’re not a robot. Peo­ple have too much trust when it comes to pro­vid­ing data. Amer­i­cans are very cu­ri­ous about their ge­neal­ogy and they take ge­netic tests that then go on­line. What hap­pens when you freely make your DNA code pub­lic? Your preva­lence for cer­tain ill­nesses can be known, and so you are giv­ing a lot of in­for­ma­tion away for free that’s very valu­able to in­sur­ers. Peo­ple have to think about what they make pub­lic on­line. And which re­la­tion­ships they want with AI.
What do you think about se­ries like ‘Black Mir­ror’ or the film ‘Her’?
Lit­er­a­ture and sci-fi help us ex­pe­ri­ence some­thing with­out ac­tu­ally ex­pe­ri­enc­ing it. And in ethics it helps us recog­nise an ex­pe­ri­ence when it hap­pens. In ’Her’, the lead is in love with an op­er­at­ing sys­tem, but it is a ro­man­tic and pla­tonic love be­cause there is no body. But in real re­la­tion­ships, the im­por­tance is in the phys­i­cal body. The ma­chine can sim­u­late but it can’t ever be con­scious, with mys­tery and emo­tions. Ma­chines should re­main as ser­vants. The day we begin to an­thro­po­mor­phise them is when we’ll begin to have prob­lems. Just look at the names: Siri, Alexia, Sofia... all of them are women’s names.
Is that sex­ism?
When we an­thro­po­mor­phise them, we make them young and docile. Big in­dus­try sees the po­ten­tial in the af­fec­tion for ma­chines. We shouldn’t cre­ate con­fu­sion about what a per­son is. An op­er­at­ing sys­tem is de­vel­oped so that when you get home it asks you how you are, how your day went, and it has sen­sors that can de­tect how you feel. That al­lows for de­vel­op­ing a re­la­tion­ship be­cause hu­mans are dri­ven by emo­tions. It’s vital not to get con­fused: it’s not real, and is pro­grammed for you to like it. And that can lead to two out­comes: mak­ing us value re­la­tion­ships with hu­mans more, or mak­ing us less in­clined to deal with com­plex human re­la­tion­ships.
And the ethics in pol­i­tics?
It is not only in Cat­alo­nia-Spain, but in the UK, the US, Brazil… the panorama is bleak. For years we have lived with po­lit­i­cal and eth­i­cal ideas that go back two cen­turies. For me, the struc­ture of po­lit­i­cal par­ties is too hi­er­ar­chi­cal and there is no democ­racy with­out a free press, and the press is not free. The re-politi­ci­sa­tion of the pub­lic is good news, but the new forms of pro­pa­ganda are not. We have to re­gen­er­ate democ­racy with fun­da­men­tal val­ues. The pri­or­i­ties are cli­mate change, AI, and bor­ders. But none of these is­sues is at the fore­front of the po­lit­i­cal agenda.

in­ter­view tech­nol­ogy

Sign in. Sign in if you are already a verified reader. I want to become verified reader. To leave comments on the website you must be a verified reader.
Note: To leave comments on the website you must be a verified reader and accept the conditions of use.

Published on

Most Viewed