Full AI Immersion

Who had the most influence on how your career has developed? 

I had the privilege of working with and learning from some of the top lawyers in the UK and globally while at HSF. If I have to name names, my most formative influences at the firm were my trainee supervisors. They were particularly important given the role they played in the formative years of my legal development. 

Chris Bushell in Litigation and Veronica Roberts in Competition both demonstrated an incredible ability to absorb a huge volume of complex information, distil it down to the essential facts and get to the heart of the legal issue. James Robinson, whom I sat with in the Tokyo Corporate team, was amazing at building deep relationships with clients, putting them at ease and establishing a lasting trust. David Coulling, in the TMT team, taught me the importance of deeply understanding the client’s business and embracing technical details. While many lawyers avoid non-legal complexities, he would use these to ask insightful questions and find novel solutions. After qualifying into the London TMT team, I was fortunate to work closely with David for several years as an associate, and he remains a valued friend and mentor. 

As you look back on your time with HSF, what do you think you gained? 

My training at HSF was world-class. The TMT team is consistently ranked Tier 1 in multiple jurisdictions, which was a key factor in Google recruiting me. The calibre of my colleagues and the matters I worked on also laid a strong foundation for my various roles at the company over the years. 

As an in-house lawyer, I’ve had the opportunity to instruct many leading firms around the world, and it has highlighted that HSF really is one of the best. Most recently, I have had the pleasure of working with Miriam Everett and Duc Tran on a pioneering global data collection effort for training complex ML models, and both are leaders in their field. 

Why did you move to Google? 

I moved to Google nearly 10 years ago because it was an opportunity to work on groundbreaking, global products at one of the most innovative technology companies of our generation. 

I joined the UK technology team and my first role was commercial legal lead for payments, which included managing the roll-out of Android Pay (now Google Pay) across Europe. This was an exciting opportunity early in my legal career, as it was one of Google’s first forays into mobile payments and the finance sector, and the product is now used by millions of people around the world. Working in what was then a small in-house team also meant I was given a lot of responsibility. In my first year, I also led negotiations on two crucial global payment processing deals covering over US$3 billion in annual sales. 

Another motivation for working at Google is the opportunity to work in different parts of the business. I became the sole lawyer in San Francisco for Area 120, Google’s startup incubator. This was a fascinating role as I became de facto general counsel for more than 40 start-ups over my five years in the team. I advised on a range of products across multiple sectors, including social media, e-commerce, digital health, wireless services, blockchain, gaming – and of course, AI. 

I was also appointed to the incubator’s investment team - a rare opportunity for a technology lawyer. In this capacity, I reviewed hundreds of applications, sat in dozens of pitch meetings and advised on funding decisions. This gave me a foundational understanding of business drivers and the challenges of launching a product to market.

You now advise exclusively on AI – what does that entail exactly?

As Senior Counsel for AI in Google Research, my primary role involves advising the AI researchers, engineers and scientists as they develop and launch new foundational models and technologies. This covers a range of tasks, including advising on compliance with the rapidly expanding legal landscape, developing internal policies on the safe and responsible development of AI, drafting external terms for our user-facing products, and engaging with regulators and other third parties. 

One thing that makes this position particularly interesting is working hand-in-hand with these experts to design the models from their inception. This includes typical AI legal issues such as procuring appropriate training data and implementing data privacy protections, but can also cover more technical topics such as the merits and risks of using a specific model architecture or loss function. Being involved from the outset means I am an integral part of the product team - again something that can be rare for a lawyer. 

What have you learned from being immersed in these businesses? 

One of the key business lessons I’ve learned from working at Google is that the focus must be on the end user and solving a problem. This is particularly relevant to AI. There is clearly a lot of hype around its potential, and large language models (LLMs) in particular have gained a lot of attention. However, one challenge now is finding real-world solutions and applications of AI that genuinely help users. 

One of the most rewarding real-world applications I have advised on is the development and open source release of Project Guideline. This digital tool leverages complex ML models to enable people who are blind and low-vision to walk or run outdoors independently, using their phone’s camera to provide real-time audio feedback. At Google Research, we have the freedom to explore a range of potential AI applications, and it’s fascinating to see what is becoming possible. 

How do you see AI and AI’s applications developing? 

It’s hard to say for sure where AI will go, but I’m excited by its ability to increase access to information and democratise learning. For example, in 2022, I led the global launch of Aloud, which harnesses advances in ML to translate and dub videos into multiple languages. This is particularly important in countries lacking access to native language content. I also led the launch of a language learning tool in which English learners can speak into their phone’s microphone, interact with an AI-powered language tutor and receive real-time, personalised feedback.

What excites you – and worries you - about AI? 

Working at Google, a huge focus for us is ensuring that AI is developed and implemented safely and responsibly. The incredible generative capabilities of AI tools, combined with the pace at which misinformation can now spread across social media, means AI companies must proceed carefully and thoughtfully. 

At the same time, AI is demonstrating an incredible ability to analyse information and make deductive leaps that are challenging how humans think across a range of disciplines. For me, this is one of the most exhilarating elements of working in AI, i.e. the unforeseen possibilities that are already emerging. At Google, we have seen several such discoveries in recent years - from AlphaGo’s famous “Move 37” to groundbreaking developments in chemistry and biology. 

How should law firms best utilise AI? 

That’s a good question and one I think about frequently. Last year, I was appointed to the Law Society’s Technology and Law Committee, which helps lawyers adopt new technologies, including AI. I am also a co-author for IAPP’s AI Governance Professional (AIGP) course, which includes guidance on how businesses and practitioners should think about implementing AI safely and responsibly. 

The legal profession has been notoriously slow in adopting new technologies. However, the sector is ripe for AI development. There is a large amount of high-quality, curated data to train models. LLMs are, by their nature, adept at textual interpretation and generation, and AI more generally is well-suited to identifying patterns and anomalies in data. As such, AI is already being used to help with traditionally repetitive and time-intensive tasks that often fall on junior lawyers, such as contract reviews and due diligence. 

Specifically, what should law firms be doing? 

I think there are several questions every law firm should be considering. The first – and one that probably doesn’t get enough attention - is how professional legal ethics apply to the use of AI, particularly generative AI. For example, can confidential client information be used to train AI models, even if they remain proprietary to the law firm? Can those models be used to generate advice for other clients? What happens if the client leaves the firm - does the model need to be retrained without their data? There are also interesting issues that go to the core of the law firm’s traditional time-based business model, e.g. what is a fair and appropriate method to bill a client if much of the work was carried out by AI. 

One issue that does receive a lot of attention is of course “hallucinations” (which are, broadly, inaccurate or misleading AI outputs). Until more robust models are developed, it seems that certain AI-generated work – especially in the legal context where the accuracy of advice is paramount – needs to be manually reviewed. In the AI sector, we refer to this as putting a “human in the loop”. 

Despite these challenges, it’s promising to see the legal profession embrace AI, and I hope it heralds an increasingly open approach to using technology in the industry.