Thank you, Hannes. It’s great to be here. The cre­ation of arti­fi­cial life is a very pop­u­lar motive in lit­er­a­ture and sci­ence fic­tion. And it’s a sto­ry that nev­er real­ly has a hap­py end­ing. The cre­ator, even though he might have the best inten­tions, ends up sum­mon­ing spir­its that he can’t control.

When we talk about tech­nolo­gies such as AI, and pol­i­cy, one of the main prob­lems is that tech­no­log­i­cal advance­ment is fast, and pol­i­cy and democ­ra­cy is a very very slow process. And that could be poten­tial­ly a very big prob­lem if we think that AI could be poten­tial­ly dan­ger­ous. Now, we heard before from Alex Lebrun that this is all sci­ence fic­tion, this is not going to be hap­pen­ing. But there are oth­ers who disagree. 

The devel­op­ment of full arti­fi­cial intel­li­gence could spell the end of the human race.
Stephen Hawking, Stephen Hawking warns arti­fi­cial intel­li­gence could end mankind”

Like this guy, Stephen Hawking, who says it could spell the end of the human race.

Nuclear pow­er gave us access to the almost unlim­it­ed ener­gy stored in an atom, but unfor­tu­nate­ly, the first thing we did was cre­ate an atom bomb. […] AI is going to go the same way.
Stuart Russel, Science Friday, April 10, 2015, The Future of Artificial Intelligence”

Or Stuart Russell, who com­pares it to nuclear tech­nol­o­gy, where first the idea was to find the ener­gy source but the first thing that was actu­al­ly made was a bomb.

Now, this may all be sci­ence fic­tion, but if you have sci­ence fic­tion you also have an idea or a vision of how the future could be. And one of these ideas is usu­al­ly this dystopia, as you can see here, the rise of the machines. 

But there’s also a more utopi­an idea like in Star Trek, where tech­nol­o­gy can meet all mate­r­i­al needs of mankind and there­fore there’s no more pover­ty or greed or hunger, and no more war. And if we talk about tech­nol­o­gy, we always have to think about what kind of future we want to have.

Now I will talk about some of the issues and chal­lenges that we face in this fourth indus­tri­al revolution. 

There’s a study by Oxford University that pre­dicts that 50% of all American jobs will be at risk in the next twen­ty years due to automa­tion. And in the past, automa­tion was usu­al­ly some­thing that con­cerned most­ly blue col­lar jobs. But it is entire­ly pos­si­ble with AI and oth­er tech­nol­o­gy that more highly-skilled jobs could be at risk.

Now, this could be a good thing. It could be could be an ally. It could help us and free us from bor­ing tasks that we don’t real­ly like to do. Or it could be main­ly a good thing for employ­ers because they have a dream work­force who nev­er com­plains, who works all the time, does­n’t have to go to the bath­room, or does­n’t join a union.

So, the ques­tion is who’s going to prof­it from all these devel­op­ments? And if we ask this ques­tion, we have to a look at who’s invest­ing in AI research. And now you course have uni­ver­si­ties, you have public/private part­ner­ships. But you also have the tech giants that invest a lot of mon­ey in these tech­nolo­gies. And you also have, of course, the mil­i­tary that invests.

Now, even if arti­fi­cial intel­li­gence won’t be as intel­li­gent as in the sci­ence fic­tion movie, even if it’s just a lit­tle self-learning and self-improving, if we imag­ine an Internet of Things where we are sur­round­ed with self-improving machines, this ques­tion about who is liable and who is respon­si­ble if some­thing happens…we will have to deal with this ques­tion. Because if you have a self-driving car and there’s an acci­dent, who’s respon­si­ble for this? Is it is the own­er, like with a pet the own­er is respon­si­ble? Or is it the man­u­fac­tur­er? Or maybe it’s the machine itself.

When we talk about arti­fi­cial intel­li­gence as not just pro­gram­ming but some­thing that is learn­ing and is devel­op­ing, this might be some­thing we would have to think about. That these machines should learn some rules, some norms, some val­ues, as well. I’m sure you all know Isaac Asimov and his Robot Laws. So they should have some sort of rules that they should­n’t harm peo­ple or should­n’t harm themselves.

Min Li Marti Ai Policy 04h39 07

If you take anoth­er pic­ture from sci­ence fic­tion you have Commander Data from Star Trek, who has an eth­i­cal sub­rou­tine [that] worked very well for him. He was prob­a­bly more human and more moral than all his col­leagues. But then again, he also had an evil twin brother.

If we talk about all these chal­lenges, we should also talk about pos­si­ble solu­tions. What can be done. What should be done. Is there any­thing to be done at all? The first idea would be to reg­u­late, maybe cre­ate a reg­u­la­to­ry agency. Maybe there’s a need for a law, or maybe just a code of con­duct. Maybe we should just ban things that we don’t want, like the Geneva Protocol that bans chem­i­cal or bio­log­i­cal war­fare. We could ban arti­fi­cial intel­li­gence in warfare.

Maybe it’s just that research should always con­sid­er eth­i­cal impli­ca­tions, like in the life sci­ences. In med­i­cine and genet­ics, that’s always a part. You have the Hippocratic Oath, you ethics boards. So maybe that should be a part of research on arti­fi­cial intel­li­gence and arti­fi­cial life, as well.

As I men­tioned before, who’s going to prof­it? So, there is prob­a­bly in each to have research that is not for prof­it. One ini­tia­tive is OpenAI, that Elon Musk and oth­ers are fund­ing, but that could also be con­ven­tion­al uni­ver­si­ty research, also on the impli­ca­tions on society.

Now, this is a high­ly high­ly con­tro­ver­sial ques­tion that we in Switzerland will vote on in June. It’s an uncon­di­tion­al basic income for all. This sounds not very real­is­tic, but if these pre­dic­tions that 50% of all jobs will be lost, then maybe we have to think about what to do with all these peo­ple who don’t have jobs any­more, and maybe an uncon­di­tion­al income could be a pos­si­ble solution.

Or we could think about what sets us apart, what sets human intel­li­gence apart from arti­fi­cial intel­li­gence. And that would prob­a­bly be cre­ativ­i­ty, social skills, empa­thy, things like that. And maybe our edu­ca­tion should focus more on these skills, so we can strength­ened the skills that set us apart.

In the end, the ques­tion is just, if we have a pow­er­ful tech­nol­o­gy, who’s going to ben­e­fit from this tech­nol­o­gy? Is it going to be the wealthy? The big com­pa­nies? Is it going to be the mil­i­tary? Or can this tech­nol­o­gy be used to solve the big prob­lems we have, like cli­mate change? Or dis­eases, or pover­ty? And I think we should make sure that AI tech­nol­o­gy, that this pow­er­ful tech­nol­o­gy, should ben­e­fit every­one and not just the few. And that is pos­si­ble, but we have to start the con­ver­sa­tion now, and we have to start dis­cussing pos­si­ble solutions. 

Thank you.

Further Reference

Artificial Intelligence, Technology with­out Alternative?, at the Lift Conference 2016 site.

This pre­sen­ta­tion, with com­plete sites, is avail­able at Klewel.