Kasparov talks calculated odds, AI, and cybersecurity (Q&A)

LAS VEGAS—Nevermind whether Han Solo, or his alien rival Greedo, shot first. The more important question the famous Star Wars scene raises, chess grandmaster Garry Kasparov asserts, is: Was Han right about the odds?

As Han steers the Millennium Falcon into an asteroid field, the Rebels’ robotic companion C-3PO warns him, using math, that his actions are likely to get them all killed. (Except maybe Threepio and his robot mechanic companion R2-D2, who don’t need oxygen to survive in the vacuum of space.)

“Sir, the possibility of successfully navigating an asteroid field is approximately 3,720 to 1,” Threepio exclaims in his tinny voice.

“Never tell me the odds!” Han retorts.

That exchange, first released in 1977 and then updated at director George Lucas’ request for a 1997 Special Edition, says Kasparov, accurately foreshadows much more about our potential relationships—and challenges—with computers and artificial intelligence than it might appear at first glance.

“Machines that know the odds will not be giving you the right answer because there are many other factors that could simply outweigh the precise calculation,” Kasparov tells me here, as he prepares to give a keynote address at the 25th annual DefCon hacker conference. Such factors, he says, might include the importance of buying a child a birthday gift, even if it breaks your personal budget, or reading subtle tics in your poker opponent’s face as a bluff. Garry Kasparov is a “security ambassador” for Avast, which sponsors this site.

This tension between the advanced pattern recognition abilities of modern AI and the value of human analysis lie at the heart of Kasparov’s presentation to the hackers gathered in casino-rich Las Vegas—an epicenter of calculated odds, often against machines. But what he wants to see is not a war against computers, or an extreme dependence on them.

Instead, he says, we should be fostering a deepening human-computer partnership.

Some call AI the future of cybersecurity. Others, including Tesla and SpaceX honcho Elon Musk, fear that it might present the “biggest risk we face as a civilization.” There is clear disagreement across, and at the highest levels of, the tech industry.

“If you have humans playing each other, the weaker player with a machine and superior process will beat a much stronger player with a machine and inferior process.”

Kasparov is no stranger to artificial intelligence. Following his famous defeat in 1997 to IBM’s chess-playing supercomputer Deep Blue, he began thinking about the intersection of human and computer intelligence. With a clear reference to—yet human distinction from—a form of AI called deep learning, Kasparov has published a new book on the subject called Deep Thinking.

Kasparov has also been thinking deeply about human rights. As leader of the Human Rights Foundation, a nonpartisan, nonprofit organization that advocates for democracy and freedom, since 2012, he published the 2016 book Winter is Coming: Why Vladimir Putin and the Enemies of the Free World Must Be Stopped.

As you might be able to guess from its title, Kasparov is no fan of the Russian president, whom he ran against in 2008 for Russia’s highest office. Nor is he fond of U.S. President Donald Trump, despite identifying as a noncitizen Republican. In columns in the Weekly Standard and New York Daily News, he has decried Trump’s affection for his longtime rival.

What follows is an edited transcript of our conversation.

Q: The ties between chess and artificial intelligence are not readily apparent, but you suggest that changed when people began building computers to play chess. Can you walk us through the connection?

People have thought that chess is the ultimate test of artificial intelligence. The founding fathers of computer science—Alan Turing and Claude Shannon and Norbert Wiener—didn’t just think about chess being solved, or not solved, by a computer, but rather about it being mastered by the machine through an AI process. Other games, they thought, were just too vulnerable for brute force.

Brute force is basically calculating. It’s also some algorithms, of course—alpha-beta then Monte Carlo—but at the end of the day, it’s minimum understanding, maximum calculation. With them, you can make summations.

Using Moore’s Law, adding more megabytes, gigabytes, terabytes, you can build a massive amount of information for playing chess. That’s helped by phenomenal processor speed, or brute force. It’s ironic, of course, because chess was considered to be such an ultimate test for artificial intelligence.

You didn’t come to this understanding of AI before your defeat by Deep Blue, right?

Regarding Deep Blue, people thought, “Oh, wow, the machine won.” I explain that in every game, whether it’s chess or Go or Shogi (Japanese chess), at the end of the day, it’s not about learning and understanding the game. It’s about beating the human opponent. The human game is too unstable to resist a powerful machine.

“While machines are destroying some jobs, they’ll create new jobs. It’s that simple.”

This is not because the machine plays a perfect game—far from that. It’s because we’ve reached the situation where any chess program that you have on your laptop is playing better than international chess masters, and any free chess app you have on your mobile phone would play better than Deep Blue.

Really?

Yes. Magnus Carlsen now is roughly ranked at 2,830. [Read more for an explanation of chess ranking.] I peaked at 2,851, Magnus’ peak was 2,872. The strength of the chess computer today is 3,300. In 1997, I could have beaten Deep Blue, if I had been better prepared, but winning one more match doesn’t change the equation. Even the best player is 500 points weaker than a computer.

But what happens when you combine a computer with a human? You match up a strong player with a computer versus another more powerful computer, and then you can see a difference. You can quantify human intuition.

The most important thing we discovered, and this is revelation that goes way beyond chess, is that while human-plus-machine beats any machine—the machine helps to eliminate the main human weakness, inconsistency—if you have humans playing each other, the weaker player with a machine and superior process will beat a much stronger player with a machine and inferior process.

How does this work outside of chess?

Let’s say the computer is doing a medical diagnosis. Whom would you trust more: a great nurse or a good doctor? The doctor might try to force the machine into his direction, while the nurse might follow the recommendation of the computer.

We cannot beat the machines. We have to join them. This is the future, and there are many problems for us to solve together.

How do you see this development impacting the job market? There’s a real fear that computers will put people out of work.

You know, Facebook has been hiring thousands of specialists because machines couldn’t identify fake news. That tells us that while machines are destroying some jobs, they’ll create new jobs. It’s that simple.

AI is going to destroy part of the world that we knew, as machines always have done in the past. The sooner it happens, the more that jobs will be created. Slowing down the process will not satisfy the demand of the market.

“America has to demonstrate what happens to those who engage in cyberattacks. Cyberwars could become a decisive factor in global politics and in bringing countries together, as the United Nations did in 1945.”

I’m 54. I’m not a part of the generation that will benefit from these new jobs, but I’m here to analyze it. I’m telling people what I saw before, in chess, and I see it happening now in security and beyond.

Machines were and are invented by humans to improve our lives, including our living standards and job efficiency. It doesn’t matter whether it is to work better with farm animals, augment human labor, or create more sophisticated technology that moves into AI.

At the end of the day, there’s still room for humans to create a territory where we will be needed and will not be replaced.

That fear of the future certainly fueled a good chunk of Donald Trump’s campaign—and now it’s fueling his presidency. We last spoke about Trump and his affection for Putin when you wrote a column for The Parallax earlier this year, just as he was being sworn in. What’s your take on the Trump-Putin relationship now, seven months into his presidency?

It’s like hiring someone who actually robbed your house to improve your security system. Putin is a dictator, and I cannot judge how much Russia was involved in fixing Trump’s victory. There was definitely help. Was it decisive? I’d rather say yes, and I think that the help was more than substantial, but again, we cannot quantify it.

Of course, I believe that Russia had compromised a man inside the group, Michael Flynn. Thank God he’s out, because with him, we would be seeing a very different world. A “grand bargain,” maybe, in Crimea, dividing the worlds and playing with Stalin’s rules.

Without Michael Flynn, and with McMaster and Mattis, Trump is definitely limited in his crazy movements. Still, at the end of the day, Putin has an agenda, and Trump doesn’t, beyond being a big man and making big deals.

It’s quite easy to make a few compliments to Trump just to play on his weaknesses. Trump benefited all his life for finding his way around rules. For Putin, it’s a match made in Heaven.

How would you define Putin’s agenda, and how does it relate to Trump?

Putin’s agenda is simple: Stay in power. Now, how you do that when your economy is in tatters, and you don’t expect any improvements? The corruption is endemic, and you understand that all your cronies are just there to take more and more money, so you need to create chaos all over the world.

Russian propaganda is based on promoting Putin as a white knight; defending Mother Russia against an evil world. To create chaos, you have to organize, and there are many opportunities now with Trump, who is trying to build this relationship, because Trump likes dictators.

Trump hates process, rules, consensus, and negotiations. He would prefer to deal with Putin than with the 28 leaders of the NATO countries. Putin is more dangerous than the Soviet regime. He built—I have to give him credit—a very sophisticated system. He created fake news on the Russian Internet more than 10 years ago, to diminish all efforts of Russian opposition. A few million dollars got him a few websites that could paralyze the efforts of his opponents. If people tried to go to the free Internet, they were easily called out by the Kremlin.

The sites produced mostly true stories that pretended to be liberal and open, but that always supported the Kremlin’s agenda. He moved this operation to Estonia, then Ukraine, throughout the Russian-speaking world. You remember the cyberattacks on Estonia in 2007? His people were also successful in building the fake-news industry and probing the weaknesses in cybersecurity.

Then he asked, why limit this operation to Russian-speaking countries? So he moved beyond. This is why I’ve been saying for years that Putin would eventually be everybody’s problem. As every dictator in the past, he will eliminate perceived enemies inside and look for enemies outside.

You’ve had personal encounters with Putin. You’ve accused him of being responsible for the murder of your friend Boris Nemtsov in 2016. What are the first steps to stop him, without having to go to war?

It took a long time for Europeans and Americans to realize that the threat of Russian interference in elections is real. As long as Putin stays in power, election interference and hacking will continue. He’s not going to leave in peace. He will make a temporary truce on one front to attack you on another.

By creating conflict, he builds an atmosphere where he can survive. Confrontation is his policy, the West is an enemy, and America is a sworn enemy No. 1. Don’t take my word for it; check out any Russian news program. He presents himself as indispensable. With living standards dropping every day, he needs confrontation as propaganda.

So, what can be done about it? I’m not a cybersecurity expert, but from what I understand about cybersecurity and cyberwar, if you are on the defensive side, you always lose.

Cybersecurity is a perfect weapon because it’s cheap. Putin doesn’t have a lot of money for massive military buildup, but hackers are cheap. He was way ahead in this game, and he can exploit the openness of Western society to promote his agenda.

The only way to win this war is basically to hit back and to make clear that any attempt will be dealt with most decisively, which former U.S. President Barack Obama failed to do. America has to demonstrate what happens to those who engage in cyberattacks. Cyberwars could become a decisive factor in global politics and in bringing countries together, as the United Nations did in 1945. We should start doing something similar—creating rules that have consequences and that must be obeyed. Putin’s Russia is the main obstacle to create any legitimate regulated space where countries can feel safe because lessons must be taught.

This sounds like you’re advocating that the U.S. “hack back” against nation-states and nation-state affiliated groups that have been attacking U.S. computer systems. Do you think that the U.S. or U.S.-allied groups should be attacking Russian infrastructure?

Yes. It should be clear that Russia has demanded this kind of attention. We have to establish the facts first, and that’s why Trump is so important for Putin. Trump muddies the water, making it harder to parse facts.

If there is to be a truce, Putin has to be forced to accept it.

Banning Kaspersky Internet Security from U.S. government offices is one of those first moves. I don’t buy that Eugene Kaspersky, someone who has been working in Russian cybersecurity for 30 years, is not connected to KGB. Eventually, Russian businesses like Kaspersky will realize that Putin is a liability.