Q&A: Sarah Barrington, PhD student studying AI harms and deepfakes

Why did you become interested in science and engineering?

I have always been fascinated in how things work and explaining the world around me. Even as a child, I spent a lot of time breaking things, trying to put things together. I was fortunate to have very supportive parents. My dad is an engineer, so supported me with tinkering with things and helping me build my own internal model of how the physical world works. I really enjoyed the sciences right up to university level, and decided I wanted to continue with what I was building then, which was an engineering skillset.

How did you get to where you are now?

At university, I was fortunate to be involved in Cambridge University Eco Racing, a racing team where you design, build and race a solar-powered car. I learned a lot about working as part of a high-performance engineering team. I also enjoyed having the opportunity to try a range of different jobs each summer, everything from sustainability consulting to finance internships. My favourite was when I got to work for McLaren as an intern in their Applied division. 

That internship turned into a real job, which I did for a few years, and I saw how these cutting-edge predictive modelling techniques that were being used in our work could be applied to even broader problems. Ultimately, this inspired me to try the world of entrepreneurship.

I worked with one of my best friends (who later became my co-founder) to build a digital marketing team and then also did some work in the blockchain space. Eventually, the blockchain side joined with a management consultancy in London and I became part of the data science team there. Ultimately, I kept finding that the parts of my job that I really enjoyed were the research-focused parts, and realised that it was time to return to academia. I knew I had to get closer to Silicon Valley and so I applied to Berkeley. I was fortunate to be awarded a Fulbright Scholarship to help me get over there and study my master’s, which turned into a PhD.

What has been your biggest achievement to date?

I think probably the biggest one would be getting into my PhD programme to work with my adviser, Professor Hany Farid, because it really was a culmination of everything that’s come before, but also a huge amount of hard work, and it just felt really right. Hearing that I’d got the internship at McLaren and receiving my offer from Cambridge were also highlights. Having said that, I don’t think I would have been able to achieve half as much in my career had it not been for my very patient and supportive family, friends, colleagues, and mentors.

What is your favourite thing about being an engineer?

I suppose I’m not a classical engineer any more – I’m more of a computer scientist. But the fundamental skillset is the same. I sit down with a challenge or a problem that often seems completely impossible to solve. I try and work back from that end goal to understand the steps to getting an answer that’s robust, trustworthy and makes sense. So, then I design experiments or tools or write code that can help get me to that answer.

I think my favourite thing about being the kind of engineer I am is the satisfaction you get when you’ve reached that end point – and you don’t always reach the end point. You have to go back to square one many, many times. But when you eventually see a project from zero to something that’s working and something that answers the question you wanted to understand, it’s the most satisfying thing in the world.

What does a typical day involve for you?

My days are very varied because as an academic, there are different types of responsibility you’ll have throughout the academic year. At the moment, I’ll wake up, probably go to campus to have a meeting with my adviser. We’ll talk about the very niche research we’ve been working on in our field of deepfake detection and AI forensics. Then I’ll go to class for a few hours. I’ll grade some papers, because I’m working as a teaching assistant. Then I might write some code where I’m building on the comments that happened in my adviser meeting. 

Then I might go and give a guest lecture somewhere talking about our work and how it’s applicable to the real world. Sometimes we also advise private sector companies or even the government to tell them about the risks and dangers of AI misuse and how our work can support their work in policymaking or in private sector innovation and development. Every day looks different. But that’s what I love about it.

What would be your advice to young people looking to pursue a career in engineering?

At school, just try and do work experience: try and shadow someone, or email local companies that you might be interested in visiting and seeing what they’re about. Even if you only spend a day somewhere, that experience is really, really valuable and will set you apart when you apply to university.

At university, it’s similar – get these internships and experiences under your belt while you’re in the safety net of studying, because the stakes are pretty low. Try different internships and summer jobs, extracurriculars as well. Get a flavour for the kinds of problems that you want to work on. 
And then beyond that for career engineers I would say staying up to date with the latest developments in tech and in engineering is really crucial to career development. Certainly, in the world of AI and computer science, there’s a new development that fundamentally changes your job every week. So being able to stay on top of that and still lead a cohesive career while adapting to these new developments is really, really important.

What’s next for you?

I spend a lot of time thinking about this. At the moment I have probably two years left of my PhD, but my entire career has been about developing methods that can go and make the world better in some way. My next step will inevitably involve taking the research from my PhD in this space – of AI harms and deepfake detection – into the world and trying to do some good with it. That might be through policymaking, or through staying in academia and trying to get on the route to professor so I can carry on research in this field and educate other students as well. But I think there is a transition that can happen from research to impact that I’m really, really interested in. After building that foundation on the research side for a couple more years, I think I’ll be going out into the world and trying to implement some of it.