Garry Kasparov, Security Ambassador of Avast and former world chess champion along with Noel Sharkey, Emeritus Professor of robotics and AI at the university of Sheffield had a great talk at the recent Viva Technology gathering. Can AI be Evil? The answer of course is no, Kasparov said “At the end of the day it’s about humans, since AI is agnostic, it’s not a magic wand nor Terminator, it’s not a harbinger of utopia or dystopia. It’s a technology and I always say humans have a monopoly on evil, so the issue is on how we choose to use the vast amount of power that will be generated by AI”.
Kasparov mentioned the question he’s usually asked: “What do you think about your defeat back on 1997 by Deep Blue? Was it a curse or a blessing?” and that eventually for him it was about timing and that it was a curse turned to a blessing, since it helped him understand that anything we do well, machines will do better. Anything that can be quantified and anything that you can qualify as a closed system (and games are closed systems with reference to chess and go), machines will prevail at.
It can be argued that the goal with autonomous vehicle behavior is to be less flawed than the behavior of a human driver. People perceive that if vehicle AI is not 100% perfect then the deal is off and they can’t accept AI on the road – that’s not a viable or realistic expectation to have from an AI or from a human driver. And that was the case with chess as Kasparov said. “The player who will win is the one who will do the least amount of errors”, so similarly, if AI makes less errors than humans in managing a vehicle and behaving properly in traffic, then it will be the best driver in all scenarios compared to the human counterpart. That will bring about less deaths, better traffic and will vastly improve out lives.
Kasparov suggested that perhaps we should change the meaning of AI to be Augmented Intelligence, in an effort to focus on the collaborative aspect of humans and machines. “Artificial” sounds alien he says, as that invokes the terrors of terminator or other “cold hearted” Hollywood creations. Even the creator of the term, John McCarthy, thought that term was a mistake on his part.
Emeritus Professor Noel Sharkey also focused on the disservice that existing AI systems do to society and the amount of wrong, racist and unethical decisions such systems make in areas like insurance, policing, loans etc, due to technological limitations or errors. For most of the cases, it looks like no one is second guessing those choices, since it’s “the system making it”. Our industry has a lot of work ahead of it and it was also mentioned that problems we consider solved, are not solved at all.
He also spoke about his work with the UN on AI warfare platforms. Russia, United States, China, Israel, Australia and the United kingdom were named as few of the players that develop autonomous weapon systems, that is, systems that can select their own targets, engage and destroy them without human intervention. He had been working for 6 years to try and stop these developments to no avail, since the idea to delegate the decision to kill, to a machine is unethical. Kasparov mentioned that complaining about the decisions machines make is like complaining in front of a mirror. Machines aggregate data and produce decisions based on our human societal problems and our own views on how to solve them. Some people try to make small changes to fix the systems and their solutions achieve nothing and are akin to distorting the mirror image to make society look better. He finally spoke on the subtle differences on semantics between deciding and choosing. Deciding is computational while choosing is human i.e. “I was told” vs “I wanted”. One thing that both speakers were unsure of, was the impact of AI on the job market on a global level, but the overall consensus was that if we work hard with AI, it can be a great benefit to humanity as an assistant, a tool.
In closing Kasparov said the following about technology and the people complaining about it: “Technology is the main reason we’re still alive to complain about technology.” and that it’s not about AI being evil, fearing it and complaining about it rather than, how us humans will choose to utilize it.
Watch the full talk on YouTube
Michael @ElecronCloud
