One-on-one | Tech

Marcin Malinowski: Artificial Intelligence is going to be a game changer for humankind

By - 17.11.2016

Google’s head of global partnerships for Eastern Europe talks AI, user privacy and how to get hired.

For many people, Artificial Intelligence (AI) is considered a distant science fiction seen in Hollywood movies such as Ex-Machina and Transcendent. But last month in front of a large audience in Prishtina, Google’s Marcin Malinowski pointed out how AI has not only been the focus of the tech world in recent times, but that within a couple of years it will be a game changer for humankind.

Talking about AI breakthroughs, Malinowski described how Google’s AI has learned to play the ancient Chinese game Go; famous for its boundless possibility of moves and the strategic thinking required to play it, Go is often considered to be more complex than chess. He highlighted that this Spring, Google’s AI managed to beat the Go world champion in what is considered to have been a historic Go battle and an ultimate victory for AI researchers.

Malinowski went on to discuss how AI can today reproduce the artworks of the greatest painters in the world. He explained how researchers from a German university developed an AI able to even understand the technique and brush strokes of painters; as a result deep learning algorithms recreated masterpieces by painters such as Van Gogh and Edvard Munch. “There is also a song performed by AI, which resembles Beatles’ music as I recall,” Malinowski told K2.0.


The endless possibilities of AI and data collection were the focus of his keynote speech during October’s fifth edition of the KOSICT Tech festival, organized by the Kosovo Association of Information and Communication Technology (STIKK) with support from Innovation Centre Kosovo (ICK) and Universum University College. Head of Google’s global partnerships for Eastern Europe, Warsaw-based Malinowski was one of the most prominent speakers at the conference, which featured talks on a number of global trending topics in IT; from digital marketing to education technology and e-sports.

K2.0 caught up with Milanowski after the conference to discuss what is required for young developers to be part of Google, his thoughts on where AI is heading and issues related to users’ privacy.

K2.0: What are the procedures for getting employed at Google? Does one have to be super-talented…?

Marcin Malinowski: Of course, yes (laughs).

… or he/she will get trained when they’re in?

These are two different topics so let me start with number one. As a manager at Google, my primary role is to develop my team. So if my team members don’t acquire new skills, and are not developing themselves and don’t get better and better in many differents systems or things that we measure, I fail — this is my failure. So learning is … [a necessity] in Google’s culture. Believe it or not, I think that I spend not less than 25 percent of my time learning. I have training of many kinds [in] products, procedures, many skills that are available.

And there are a lot of diverse people from whom you can easily learn. In Google as a culture, there are no stupid questions. Only answers can be stupid. So, we know we can ask stupid questions to anyone and nobody is going to make bad use of that. So we give the answers, we talk a lot, we communicate and try to build each other’s trust and build each other’s expertise.

Now back to the question of the hiring process. Certainly, Google is a company that attracts the highest standards. We try to hire the best people in the world. That’s the whole idea and criteria.

"Lots of products my team is working on didn’t exist two years ago — if they were not capable of growing they would be in trouble."

So we don’t hire just talent, we hire talent for certain positions; any application has to be related to an open position. We are measuring a couple of elements and attributes that are very important. One of them, is something that we call ‘general cognitive abilities’ — it is a skill how to think, how to put logical things together, and maybe assess or estimate some values out of some logic, just to see if they are logically putting things together. If they are connecting dots, that’s very important.

And of course role related knowledge is very important because general cognitive abilities are a universal value that every ‘Googler’ [Google employee] has to have, while there are a certain set of skills that are very much related to certain positions to which we are hiring.

And a very important element is also to tell if the person is able to change jobs in the future. What we are trying to asses is the ability to change position and to develop later on. Google didn’t exist 18 years ago and lots of products my team is working on didn’t exist two years ago so I was only hiring specialists in a very narrow field — if they were not capable of growing they would be in trouble.

What’s your advice for the enthusiastic developers that want to work for Google one day?

Really measure your skills and challenge yourself, before anybody else is challenging you.

Second: Be yourself. If you get an interview absolutely be yourself. Don’t pretend to be somebody else because that isn’t going to fly and remember that Google is open to every single kind of person; we embrace the diversity. We are trying to build an organization that includes everybody no matter what beliefs they have, gender they are, or sexual orientation.

I’ll tell you one story: There was a time when a new YouTube application was being built for mobile phones and of course it was uploaded to many Googlers’ phones so they could play with it for a couple of days and weeks and developers were analyzing what was happening. It turned out that something like 17 percent of movies that were uploaded through the YouTube application were uploaded upside down. And it turned out that there were 17 percent of left handed people in the tested population, which is a perfect example of how important it is to have all sorts of people so we could track issues before the product was released. That day was equally good for left and right handers and that I think proves the value of diversity in every organization, including Google.

What are the criteria for opening a Google office in a region? Are you planning to open an office here soon?

So there are a couple of elements. First of all, Google is a global company and tries to optimize operations, and one of the things for the organization is to put a lot of operations into hubs, regional hubs. For Europe the largest regional hub is Dublin, which has 5,000 people if I’m not mistaken, the second largest office is in London, which is again a couple of thousand people. And Poland happens to be the third largest location for Google, which has around 1,000 people combined in three offices.

It doesn’t always make sense to put people in local offices. A lot of operations are today handled by technology and by remote presence. And it is not economically smart to have offices everywhere, and to operate locally. A lot of our products are available remotely.


Which do you believe is the most important project that Google is working on right now?

I think DeepMind, I think that’s the most powerful project we are doing and I think that generally speaking the most interesting phenomenon happening in the world no matter if we are talking about Google or IBM… everybody is looking at this industry and this opportunity. I think it’s going to be a major game changer in the next tens of years probably. It has already started changing quite dramatically the products that we are using today, including Google search and many other products like Google Assistant and there are a lot of new things thanks to that and we are going to see a lot more things happening in the coming years.

Are there dangers of Artificial Intelligence, similarly to Hollywood movies where we see robots taking over humankind?

From what I am hearing [engineer and AI critic] Elon Musk is afraid of the turnouts. We are not. At Google we happen to think the opposite way, that this is only going to be beneficial for humankind. And that’s going to make our lives richer, rather than being a threat to us.

How does Google measure things needed for the future? For example reducing road traffic?

So there is this whole division at Google called Google X. These are the people that are responsible for the moonshots, for the big projects that are really really big. They wouldn’t consider working on a project that doesn’t make a change ten times something.

If the task was, ‘Let’s produce a car that drives twice as fast,’ they wouldn’t even bother building it. Also if somebody came to them and said, ‘Hey guys, why don’t you take the car and optimize it by changing the wheels and do stuff like that,’ there would be a whole discussion about it but they’d prefer to take a task [such as], ‘Hey guys, why don’t you rethink and resolve the issue of transportation from A to B, of 1 to 5 people, at limited cost.’ So they would rather work on projects that precisely define the goals rather than solutions.

Generally speaking, when Google X is looking at an issue, it needs to be something that tries to address something really really big. One of the examples that started at Google X was ‘Project Loon.’ The problem was how to deliver internet to another billion users, and one of the ideas that got developed was to put together the balloons that would be travelling over the stratosphere and they would bring internet down to the ground. Something one couldn’t imagine 10 years ago.

Any data that you put into the Google cloud is distributed to thousands of servers, hard drives in different locations. And the only thing in the world that can put these pieces together is your login and password.

What is Google doing regarding user privacy and criticism about censuring results?

There are very few policy rules that apply to what is searchable and what’s not. We believe that we are supposed to bring all the information of the world, making it universally accessible for people. But we are also trying to make sure that anybody could sit with his or her 7-year-old son and do the search and see content that would not offend the child. If Google ads are shown around some content, we would like to be sure that a 5- or 7-year-old could handle this. So this is one of the measures that we put in place: what we want to present as content and what not to.

It’s the same thing as what we let onto YouTube. There are certain rules that need to be obeyed; this is obvious, this is part of the deal, this is part of the platform’s policy. One could debate who are we to dictate what’s right and what’s wrong. Definitely we are not the smartest people on the planet but we are consciously trying to build platforms that are going to be useful and according to what we see and believe, these rules should apply.

Another topic is regulation that is applied on us, and on Google search results, coming from local governments, EU governments. It depends where the search is happening, where the user is sitting and sometimes where the data is processed.

Do you think that Google is protecting private data in the best way possible?

I think we are one of the best in protecting users’ data, if not the best. We handle data with extreme care. I’ll tell you how we handle your Gmail account if you have one. Any data that you put into the Google cloud is distributed to thousands of servers, hard drives in different locations. And the only thing in the world that can put these pieces together is your login and password. So there is no risk in the world that somebody steals a hard drive from any location — if somebody steals the hard drive they only get meaningless sets of bytes. Again the login and password is making the magic to put the pieces of the puzzle together and to drive data from different locations, different servers, to make logical sense out of it.

So, in the logical and physical areas I think we are ok. Then in regulatory terms, I think we are very very clear on our policies, how we handle the data, this is all public and super transparent and we are not doing anything to hurt the relationship with users. We are totally open. We provide tools to the users, in which you can check how much we know about you and what kind of knowledge we have.K

This conversation has been edited for length and clarity.

Photos: Majlinda Hoxha / K2.0.