Life at DeepMind
Meet Edgar Duéñez-Guzmán, a research engineer in our Multi-Agent Research team who uses his knowledge of game theory, computer science, and social evolution to improve how AI agents work together.
What prompted you to pursue computer science?
I wanted to save the world for as long as I can remember. That’s why I wanted to be a scientist. Although I loved superhero stories, I realized that the real superheroes were scientists. They were the ones who gave us immaculate water, medicine, and understanding our place in the universe. As a kid, I loved computers and I loved science. But growing up in Mexico, I didn’t feel like studying computer science was a viable option. So I decided to study mathematics, as a solid foundation for computer science, and I ended up writing my master’s thesis in game theory.
How did your studies influence your career?
I did biological simulations for my PhD in computer science and ended up falling in love with biology. Understanding evolution and how it shaped the Earth was stimulating. Half of my dissertation was about these biological simulations, and then I worked in academia studying the evolution of social phenomena like cooperation and altruism.
From there, I went to work in Search at Google, where I learned how to deal with huge scales of computation. Years later, I combined all three elements: game theory, the evolution of social behavior, and large-scale computing. Now I apply these elements to create AI agents that can learn to cooperate with each other and with us.
What made you decide to apply to DeepMind and not another company?
It was the mid-2010s. I had been following AI for over a decade and I knew about DeepMind and some of their successes. Then Google acquired the company and I was really excited. I wanted to join, but I lived in California and DeepMind was only hiring in London. So I kept up with the developments. As soon as the California office opened, I was first in line. I was lucky enough to be hired in the first group. I eventually moved to London to do full-time research.
What surprised you most about working at DeepMind?
How incredibly talented and amiable people are. Every person I’ve talked to has an stimulating side to them outside of work. Professional musicians, artists, super fit cyclists, people who’ve appeared in Hollywood movies, math olympiad winners – you name it, we’ve got it! And we’re all open and committed to making the world a better place.
How does your work lend a hand DeepMind make a positive impact?
The foundation of my research is creating knowledgeable agents that understand cooperation. Cooperation is key to our success as a species. We can access information from around the world and connect with friends and family on the other side of the world through cooperation. Our inability to address the catastrophic effects of climate change is a failure of cooperation, as we saw at COP26.
What is the best thing about your job?
Flexibility to pursue the ideas I consider most essential. For example, I would like to lend a hand apply our technology to better understand social problems such as discrimination. I presented this idea to a group of researchers specializing in psychology, ethics, integrity, neuroscience, and machine learning, and then created a research program to investigate how discrimination can arise from stereotypes.
How would you describe the culture at DeepMind?
DeepMind is one of those places where freedom and potential go hand in hand. We have the opportunity to pursue ideas that we think are essential, and there is a culture of open discussion. It is not uncommon to infect others with your ideas and build a team around turning them into reality.
Are you a member of any DeepMind groups? Or other activities?
I love getting involved in extracurricular activities. I am a facilitator for Allyship workshops at DeepMind, where we strive to empower participants to take action for positive change and encourage allyship in others, contributing to an inclusive and equitable workplace. I also love sharing research and talking to visiting students. I have created a publicly available educational tutorials to explain AI concepts to teenagers. The tool is used in summer schools around the world.
How can AI maximize its positive impact?
To have the greatest positive impact, the benefits simply need to be shared widely, rather than held by a few. We need to design systems that empower people and democratize access to technology.
For example, when I was working on Wavenovel Google Assistant voice, I felt it was chilly to work on technology that billions of people apply now, whether it’s Google Search or Maps. That’s nice, but then we did something better. We started using that technology to give a voice to people with degenerative diseases like ALS. There are always opportunities to do good, we just have to take advantage of them.
What are the biggest challenges facing artificial intelligence?
There are both practical and societal challenges. On the practical side, we are working tough to make our algorithms more strong and adaptable. As living beings, we take robustness and adaptability for granted. A slight rearrangement of furniture does not cause us to forget what the fridge is for. Artificial systems really struggle with this. There are some promising leads, but we still have a long way to go.
On the societal side, we need to collectively decide what kind of AI we want to create. We need to make sure that whatever is created is secure and beneficial. But that’s especially tough to do when we don’t have a perfect definition of what that means.
Which DeepMind projects do you find most inspiring?
I’m still high right now Alpha-Compositionour protein folding algorithm. I have a biology background and understand how promising protein structure prediction can be for biomedical applications. And I am particularly proud of how DeepMind has made the protein structure of every known protein in the human body available in global datasets, and has now made available nearly every catalogued protein known to science.
Do you have any tips for people who want to become DeepMinders?
Be playful, be adaptable. I couldn’t optimize my career to get to DeepMind (there wasn’t even a DeepMind to optimize for!). But I could always allow myself to dream about the potential of technology, to create knowledgeable machines and improve the world with them.
Programming is stimulating in itself, but for me it was always more of a means to an end. It allowed me to stay current with the comings and goings of technology. I wasn’t tied to the tools, I was focused on the mission. Don’t focus on the “what”, but on the “why”, and the “how” will reveal itself.