Close
After studying Economics at Cambridge and working in finance, Theo's curiosity about AI led him to UCL's AI for Sustainable Development MSc, where he found a passion for ethical AI research.
What’s your background?
I was born in Harlow, north of London. Since then, I’ve refused to move more than about 50 miles from London for longer than a couple of weeks. I guess there’s some kind of aura that always pulls me back towards it!
Initially, I studied Economics for undergrad at Cambridge, graduating in 2020. After that, I worked for three years in the financial sector (in asset management and banking roles), prior to enrolling at UCL.
Why did you choose to study Computer Science at UCL?
Around spring 2023, one of my friends told me about ChatGPT, and how it was going to steal his job. I’d never heard of it, and out of curiosity started to play around with LLMs.
Within a couple of weeks, I was so fascinated, excited and terrified about the capabilities of AI models that I started looking into Master's courses, with a view to transitioning into the AI sector.
I was attracted to the AI for Sustainable Development at UCL course due to its unique blend of:
- teaching AI from a rigorous and mathematical perspective and
- its focus on applying AI technologies to ethical use cases, like helping to tackle the climate crisis.
It all happened very quickly, but sending off that application might be the best decision I’ve ever made.
What were the highlights of your CS programme?
My answer may be biased by the fact that, to begin with, my early stages of the course were a relative fight for survival, with a series of long evenings tackling challenging coding assignments (perhaps explained by my initially not-so-expert Python knowledge); if you ask my coursemates, they probably didn’t find them as challenging!
As we moved into the second term, I enjoyed the content more and more, with a better grasp over the theory behind key concepts like deep neural networks, which are certainly a bit of a headache to digest at first.
My favourite module was Accountable, Transparent and Responsible AI, which focused on bias, fairness, explainability, and safety in AI models.
For me, this was where the lecturers’ passion for the subject (and crucial nature of their research) shone through the most. It may go without saying, but the general standard of teaching is superb.
What were the highlights of your time generally at UCL?
Without a shadow of doubt, my coursemates made the course. We were a relatively small cohort, but all from different backgrounds – 15+ countries, some had just finished undergrad, some from industry, some from other lines of academia.
Everyone was united by a desire to learn, and the lecture environments were the most interactive I’ve ever witnessed – there were about 20 brilliant questions per lecture (I’ll only claim responsibility for a couple across the year).
Each person had such interesting stories, opinions and perspectives, but above all, contributed to a genuine culture of kindness, compassion and support. This bound people together and helped them through academic and personal challenges.
What industry and career opportunities did you get access to on your programme?
In my case, the majority of industry exposure came through the summer project / dissertation.
When selecting projects, we had the opportunity to collaborate with over 20 industry partners. One of these was a firm called Holistic AI, founded approximately 5 years ago by UCL alumni, which operates in the field of AI governance.
I had an initial interview with a very talented individual (who happened to do the same course as me the year prior, I’m led to believe with partially greater success) and was immediately sold on the company mission, which aligned with my views on the need to empower organisations to develop AI in a safe and transparent manner.
When working on the project over the summer, I had the pleasure to meet and interact with people across the company, and I now work there full-time.
What did you write your dissertation on?
The title of my thesis was ‘Advancing Stereotype Auditing in Large Language Models’. In essence, this involved developing a scalable framework for talking to LLMs (such as GPT) and rigorously evaluating whether they have tendency to generate harmful biases, such as gender, racial, or LGBTQ+ stereotypes, in their outputs.
This evolved into exciting research that was published to two workshops for NeurIPS, which is a leading conference in the machine learning field.
As much as it would be tempting to claim all the credit, the guidance and expertise from my future colleagues at Holistic AI was the key in making this happen. They have an incredibly impressive record at helping Master's students transform summer projects into published research, and would be an excellent project partner for future students.
How did you manage your workload, and what advice would you give to prospective students in terms of being successful and making the most of the opportunities at UCL?
Hopefully, the department doesn’t slaughter me for this statement, but my general advice to manage the workload is to try not to work all the time!
The UCL teaching staff pay close attention to workload, and schedule the assignments to ensure deadlines overlap as little as possible, but it is still tempting to just live from assignment to assignment.
If you schedule your workload properly, you will inevitably have a lot of free time and the year will quickly fly by, so make the most of everything UCL has to offer – you’re unlikely to get another chance to interact with so many inspiring people over such a short period of time, in London!
UCL has about 10 million clubs (I think the AI Society is great but there are other ‘gems’ like the Taylor Swift Society, apparently, if the former isn’t your thing), which are a wonderful way to interact with the wider student body.
Where are you working now, and what are your career goals? How did your programme help you achieve this?
After working with Holistic AI on my summer project, I was offered the opportunity to return there full-time as an AI Safety Researcher. The focus of my work remains on LLM auditing and evaluation, from both a research and client implementation standpoint, which aligns closely with the work we have collaborated on over the past year.
I was initially introduced to the company through the AI for Sustainable Development programme, so choosing to study at UCL has had a clear and immediate impact on my career.
My experience at the company has been excellent – the people are fiercely intelligent, driven and collaborative. I plan to pursue a long career in ethical AI research, and am very grateful to have found an industry in which I feel genuine passion, having lacked this in hindsight in my previous career (financial sector).
I perhaps retain a lingering ambition to move into politics one day – the self-anointed ‘AI and Economics Expert’ who could deliver the change we need to revolutionise the UK’s ailing economy…
Why would you recommend this course?
I could name a very long list of reasons to recommend the course, but the first three would be (no particular order to avoid upsetting anyone):
- fellow students
- industry partners and direct job opportunities
- world class teaching / project supervisors.
I was nervous to take the plunge from working in a relatively stable industry to gingo back to study in a completely new domain, but each of these components has made the transition much easier and equipped me with the tools to pursue a career in the AI sector, which was my ultimate objective when enrolling onto the course.