Robots, AI and podcasting: a Q&A with Pieter Abbeel

Pieter Abbeel, professor of electrical engineering and computer sciences, is already well-known as a leading researcher of robotics and artificial intelligence. But in the spring of 2021, he added podcast host to his accomplishments with the launch of his popular show, “The Robot Brains Podcast.” In each episode, Abbeel sits down with top experts and entrepreneurs to discuss the real-world applications of AI, along with its many possibilities. Now in its second season, the podcast is available through all major providers, including Apple, Google, Spotify and YouTube.

We asked Abbeel to tell us more about his experience with podcasting and how it has shaped his own thinking about communicating AI to a broader audience.

Why do a podcast?

I love teaching and research — both are a lot of fun — so I’ve put many of my classes online. I see the podcast as another way to educate people in an enticing way. It’d be great if our alumni came back to school and kept taking classes. But most people don’t have the schedule or excitement to go back to school, so if they can be entertained and learn at the same time, why not?

What’s possible in AI and robotics today is so different from what was possible even five years ago. It’s a space where things are changing very fast and in ways that are going to impact people’s lives. My podcasts highlight the people who are making it happen. Getting into where those pioneers are coming from, what they’re excited about, what drives them, what they did earlier in their lives — it’s really fun to share that.

Who is this podcast for?

One audience is AI researchers, who often know the guests on the podcast, and so it’s another way to get to know other researchers in the field and get their views. At the same time, and very importantly, the podcast is set up to be accessible to anyone. We try not to assume any prior knowledge of our audience, and we always get the guests to explain all the basic concepts and the impact in the real world, making sure that there are no assumptions.

How did you get started?

In my lab at Berkeley, especially during the first year of the pandemic, I was thinking that there’s only so much we can do. It was very limiting so, to mix things up, I had some researchers join in for a conversation with me via Zoom. That way, students got to see something different and learn more about the research process and the people behind the research, rather than about a specific research result. The students loved it, I loved it and the guests loved it — it seemed good. So I thought, if I’m going to have conversations with these famous researchers, why not record it to let everyone listen, instead of just my 10–20 students who are in the lab?

How do you decide on the topics and guests?

I’d say about half of our guests are researchers and half are entrepreneurs. We have a set of topics that we’re looking to cover, but the guests can take them anywhere they want. Whatever subject a guest is most excited to talk about, it’s probably what our audience wants to know more about. Some guests are very famous — they might have started a company that’s a unicorn or they’re doing well-known things in the world — and that draws people in. But we actually try to do it in a way where about half the guests are not as well-known by a broader audience. We want to mix that up, because my belief is that the learning that happens is often higher when listening to the less famous guests because the audience hasn’t already heard them somewhere else.

What are you hoping the podcast listeners will learn about robotics and AI?

I think there are many headlines that are quite impressive about AI. Some carry a lot of truth, some less so. But the goal really is, if people listen to the podcast, they’ll get an understanding of what AI can do, where is it in our lives and what it will become. That’s the biggest part. For example, we’ve looked at how AI is helping with blood diagnostics, so you can have blood diagnostics at home. Then there are drone deliveries — which are actually already happening in Africa, where the road network is slower than it is here, and they’re doing blood supply deliveries with drones. Another example is how AI can help us understand how to fight climate change. There are all of these things that we care about, and AI could play a role to help.

We also try to give people an understanding of how AI works. It’s not if-then programming. Rather, AI is based on learning from data. There are neural networks, inspired by the brain, that internalize patterns from data and then become pattern recognizers. So we dive into those things in a way that allows people who might not have studied computer science to get a sense of how it works.

How has working on this podcast influenced your own views about education?

I’ve come to think of the podcast as continuing education. Guests are excited about the episodes, and audiences are excited about learning through it. So, in some sense, we’re providing a form of entertainment, even though I wasn’t really thinking that way initially. But as I look at our audiences and realize who listens and what people pick up from it, I realized that maybe this is actually the best way to do continuing education. If I have to choose between putting another course online or spending more time in the podcast, maybe it’s the podcast that will educate more people.