Life in DeepMind
Meet Edgar Duéñez-Guzmán, a research engineer in our Multi-Agent Research team, who draws on knowledge of game theory, computer science, and social evolution to make AI agents work better together.
What got you into computer science?
I’ve wanted to save the world ever since I can remember. That’s why I wanted to be a scientist. While I loved superhero stories, I realized that scientists are the real superheroes. They are the ones who give us clean water, medicine and an understanding of our place in the universe. As a kid I loved computers and I loved science. Growing up in Mexico, however, I didn’t feel that studying computer science was possible. So I decided to study mathematics, treating it as a solid foundation for computing, and ended up doing my university thesis on game theory.
How did your studies affect your career?
As part of my PhD in computer science, I created biological simulations and ended up falling in love with biology. Understanding evolution and how it shaped the Earth was fascinating. Half of my dissertation was on these biological simulations, and I went on to work in academia studying the evolution of social phenomena such as cooperation and altruism.
From there I started working at Google Search, where I learned to deal with huge scales of computation. Years later, I put all three pieces together: game theory, the evolution of social behaviors, and large-scale computation. Now I use these pieces to create artificially intelligent agents that can learn to cooperate with each other and with us.
What made you decide to apply to DeepMind over other companies?
It was the mid-2010s. I had been following AI for over a decade and knew about DeepMind and some of their successes. Then Google acquired it and I got really excited. I wanted in, but I lived in California and DeepMind only hired in London. So, I continued to monitor the progress. As soon as an office opened in California, I was first in line. I was lucky enough to be recruited into the first team. Eventually, I moved to London to pursue research full-time.
What surprised you the most about working at DeepMind?
How ridiculously talented and friendly people are. Every person I’ve spoken to also has an exciting side outside of work. Professional musicians, artists, super fit cyclists, people who have appeared in Hollywood movies, math Olympiad winners – you name it, we’ve got it! And we are all open and committed to making the world a better place.
How does your work help DeepMind make a positive impact?
At the core of my research is the creation of intelligent agents that understand cooperation. Collaboration is the key to our success as a species. We can access the world’s information and connect with friends and family on the other side of the world thanks to collaboration. Our failure to address the devastating effects of climate change is a failure to cooperate, as we saw during COP26.
What is the best thing about your job?
The flexibility to pursue the ideas I believe are most important. For example, I would love to help use our technology to better understand social problems such as discrimination. I pitched this idea to a group of researchers with expertise in psychology, ethics, justice, neuroscience, and machine learning, and then created a research program to study how discrimination might stem from stereotypes.
How would you describe the culture at DeepMind?
DeepMind is one of those places where freedom and possibility go hand in hand. We have the opportunity to pursue ideas that we believe are important and there is a culture of open discourse. It is not uncommon to infect others with your ideas and form a team around making them a reality.
Are you part of a team at DeepMind? Or other activities?
I like to participate in extracurriculars. I am the facilitator of the Allyship Workshops at DeepMind, where we aim to empower participants to take action for positive change and encourage allyship in others, contributing to an inclusive and equitable workplace. I also enjoy making research more accessible and talking to visiting students. I have created publicly available training schools for explaining artificial intelligence concepts to teenagers, which have been used in summer schools around the world.
How can AI maximize its positive impact?
To have the most positive impact, it simply needs to be that the benefits are widely shared, rather than retained by a small number of people. We should design systems that empower people and that democratize access to technology.
For example, when I was working WaveNet, the new voice of Google Assistant, I felt it was great to work on a technology that is now used by billions of people, in Google Search or Maps. It’s nice, but then we did something better. We started using this technology to give people with degenerative disorders like ALS their voice back. 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 social challenges. On the practical side, we are hard at work trying to make our algorithms more robust and adaptable. As living creatures, we take robustness and adaptability for granted. The slight change in the arrangement of the furniture does not make us forget what the refrigerator is for. Artificial systems really struggle with this. There are some promising leads, but we still have a way to go.
On the social side, we need to collectively decide what kind of AI we want to create. We need to make sure that what is being made is safe and beneficial. But this is especially difficult to achieve when we do not have a perfect definition of what this means.
Which DeepMind projects do you find most inspiring?
Right now I’m still riding his high AlphaFold, our protein folding algorithm. I have a background in biology and understand how promising protein structure prediction can be for biomedical applications. And I’m especially proud of how DeepMind released the protein structure of all known proteins in the human body to the global datasets, and now released almost all proteins known to science.
Any tips for aspiring DeepMinders?
Be playful, be flexible. I couldn’t optimize for a career leading to DeepMind (there wasn’t even DeepMind for optimization!) But what I could do was always allow myself to dream about the possibilities of technology, building intelligent machines and improving the world with them.
Programming is exciting in its own right, but for me it’s always been more of a means to an end. This enabled me to stay current as technologies came and went. I wasn’t tied to the tools, I was focused on the mission. Don’t focus on the “what”, but the “why” and “how” it will manifest.