Tuomas Sandholm applies game theory in his companies and encourages everyone to study AI

Alumnus of the Year 2019 at Aalto University School of Science, Tuomas Sandholm, was honored in a very distinguished way this autumn. US investment bank Goldman Sachs named him one of the 100 Most Intriguing Entrepreneurs of 2020.

In addition to his work as a Professor of Computer Science and researcher of artificial intelligence, Sandholm is a long-term serial entrepreneur. The tribute from Goldman Sachs was awarded for his work on a company called Strategy Robot. The startup founded in 2018 uses applied game theory, artificial intelligence (AI) and optimization, for example, to assist the United States Department of Defense.

In 2017, several US media wrote about the Libratus poker bot developed by Sandholm and his colleagues that managed to beat four professional poker players at a two-player game of Texas Hold ‘Em. The group of researchers led by Sandholm developed the bot in order to test automated decision-making based on a game theory. In 2019, his bot Pluribus won the best professional players in Texas Hold ‘Em with several players. According to Sandhold, it was the first AI-based system that performed better than humans in a game that wasn’t a zero-sum game between two players.

Sandholm founded Strategy Robot to apply similar technology for the use of the U.S. Department of Defense in planning and war games, for example. Media such as US magazine Wired have since written about the company.

Sandholm believes that the honorary mention by Goldman Sachs is due to the fact that the list includes entrepreneurs and companies working with something new and revolutionary. ‘It’s nice to have a leading instance like Goldman Sachs note us already at this stage,’ he says.

At the beginning of December, it was announced that Sandholm wins also the 2021 Robert S. Engelmore Memorial Lecture Award from the Association for the Advancement of Artificial Intelligence (AAAI) for his AI research and service to the AI community. He will receive the award at AAAI's Innovative Applications of Artificial Intelligence conference in February 2021.

‘I have always found it important to make real-life applications as well’

Selecting Sandholm as the Alumnus of the Year 2019 at Aalto University School of Science was based on the fact that he has successfully commercialized his research and translated it into innovations and business in his academic career.

‘I have always found it important that academic work has an impact in the real world and the world becomes a better place through this work. I have carried out artificial intelligence research since 1989, and I have always found it important to make real-life applications as well instead of just theory, although my research group does both.’

Even before founding Strategy Robot, Sandholm had worked as an entrepreneur for two decades.

He developed online trading centers based on optimization as early as the late 1980s and founded his first company CombiNet in 1997. It managed to arrange over 800 combinatorial multi-attribute auctions where the buyer acquired combinations of items instead of single ones, with over $60 billion in total spend and over $6 billion in generated savings, before the company was sold in 2010.

The next company was Optimized Market. It sells software for companies in the advertising industry to help with the timing, pricing and distribution of advertising campaigns. The objective is to improve the advertising chain and make advertisements visible to the people actually interested in them.

There is yet another company utilizing game theory in the making. The company, still in its early stage, is called Strategic Machine and aims to help private companies instead of operators in the public sector.

‘Business strategies today certainly don’t optimize strategies in a game-theoretical way but are based on heuristics. Industries, where for example pricing and product portfolio are optimized, do not consider the answers of competitors. This means that there is quite a lot of work in terms of making things better and in a completely new way,’ Sandholm says.

Industrial engineering and artificial intelligence

Sandholm studied industrial engineering in the 1980s at the Helsinki University of Technology, which later became part of the current Aalto University. He liked being in Otaniemi and made friendships that have even lasted until this day.

Already back then, Sandholm had an interest in combining business economics with computer science. In his studies, he delved into corporate strategy, international marketing, system analysis and operational research as well as artificial intelligence.

The technicality of applied mathematics courses appealed to him. ‘Later, at the end of the 1990s when the study of artificial research became internationally more mathematical, it was good that I had taken applied mathematics seriously.’

Sandholm graduated as a Master of Science in Engineering in 1991 and set off for doctoral studies in the United States immediately after that. Currently, he works as a Professor of Computer Science at the Carnegie Mellon University in Pittsburgh, Pennsylvania and publishes in the fields of artificial intelligence and applied mathematics.

Everyone benefits from studying artificial technology and applied mathematics

For current students, Sandholm’s advice is that whatever you do study, the basics need to be in order. ‘I see it as a pyramid of knowledge. If you do high-level research at the top, you need a strong foundation – the pyramid won’t hold without a broad and strong base.'

Sandholm recommends for all adolescents to get to know artificial intelligence and applied mathematics, at least on some level. ‘After ten years, there will probably not be many industries where you can carry out research without understanding artificial intelligence to some extent.’

Those interested in a doctoral thesis should focus on things that are new and relevant and boldly apply for research groups carrying out the best research in their field. ‘For students of industrial engineering and business, I would say that it’s helpful to have expertise in a certain field, not just management. A career can be more satisfying once you truly understand the field you work in.’

In terms of his own career, Sandholm hopes that he will get to witness the game theory being incorporated widely into applications and products, through Strategy Robot and Strategic Machine, for instance. He also hopes that the theory he has developed could somehow be used in the fight against climate change. ‘I’ve commissioned a couple of market studies on how our technologies could help prevent climate crisis. Yet we still haven’t found a so-called killer application.’