Pauline Rowntree

Fun fact: there are three recognised types of AI — Narrow (also called Symbolic or Weak AI), General AI (not to be confused with generative AI), and Super or Strong AI. While much of today’s technology is still rooted in Narrow AI, its applications are vast and growing. AI is no longer a niche or futuristic concept; it is deeply embedded in the world around us — from automating our homes to enabling innovative developments in research labs.

In data science, for instance, we might use complex regression techniques to deal with multi-collinearity or apply Siamese neural networks to assess the similarity between two inputs — a method particularly impactful in forensic science. Machine learning introduces additional tools such as conformal prediction and conformity scores, which help quantify how "unusual" a prediction is under the assumption of independently and identically distributed (IID) data. The point is clear: AI is everywhere, and its future — particularly in fields like Intellectual Property (IP) law — is both exciting and uncertain.

When Technology Meets the Law

During my internship week at Mewburn Ellis, I had the opportunity to reflect on both the promise and the challenges that AI brings to the legal world. A standout moment was discussing and watching the Emotional Perception AI v Comptroller-General of Patents case in the High Court — a pivotal example of the legal system grappling with how we treat AI-generated inventions. Since AI systems are built from code, mathematics, and algorithms, there is an open question about whether — or how — these outputs should be patentable. As this case and others evolve, they will set critical precedents for the future of IP protection in an AI-dominated world.

According to a 2025 Stanford University report, around 66% of AI models released in 2023 were open-source, and most large language models (LLMs) relied on web-scraped data for training. This raises complex issues around infringement, licensing, and the ownership of AI-generated content. Cases like Andersen v. Stability AI et al. and Getty v. Stable Diffusion underscore the growing volume of litigation arising from unlicensed use of copyrighted materials in training data and the uncertain legal status of AI-generated works.

From Algorithms to Inventions

Although my internship was not primarily focused on AI, coming from a core AI background — where every discussion and document revolves around it — it gave me a valuable perspective on the legal and procedural foundations that drive innovation. With a deeply technical academic background, I came to genuinely appreciate the expertise that patent attorneys bring — particularly in determining whether something genuinely qualifies as an invention. What may seem novel from an academic or engineering perspective does not always satisfy the criteria of patent law, which places equal emphasis on inventiveness, technical contribution, and, crucially, non-obviousness.

This last point was especially well-illustrated during a talk by Rebecca Frith, where she broke down the fundamentals of patentability using real-world case law. Her discussion helped illuminate how patents are less about rewarding raw creativity, and more about recognising inventions that advance the state of the art in ways that would not be obvious to someone skilled in the field. For those of us working at the intersection of law, science, and technology, this is a vital concept. As AI continues to blur the lines between human and machine ingenuity, understanding these distinctions — and how they are interpreted in legal systems — will be central to shaping the future of innovation.

One of the highlights of the internship came when we were expertly challenged — in the best way possible — by Partner Camille Terfve, who walked us through how inventor meetings are typically run. She prompted us to think critically about how attorneys extract the true inventive concept from technical discussions and how that translates into claim language. To put theory into practice, we engaged in a fascinating case study discussion on the seminal paper "Attention Is All You Need" — the foundation of the Transformer architecture powering many of today’s leading AI models. This exercise helped us understand how a high-level academic concept could be framed in terms of patentable subject matter.

Later in the week, we focused on claim amendments, led by Partners Dan Brodsky and James Leach, where we had the chance to rewrite claims based on a set of objections. What made this exercise especially valuable was how collaborative it was — we did not just passively watch; we discussed every step together, thinking through how claims should be logically structured, how to maintain clarity while preserving scope, and how to anticipate potential objections from examiners. We capped this off by working on a detailed case study with Partner Lucy Coe, where we prepared a formal response to an examination report. This included amending the claims to clarify key inventive aspects such as the use of a time-decay algorithm for real-time anomaly detection in clustered web-request data. Being part of the drafting process — especially seeing how legal arguments are crafted to distinguish from prior art — gave me an invaluable glimpse into the intellectual precision and strategy that goes into prosecuting a patent. It was a rare and rewarding experience that seamlessly combined technical insight with legal reasoning.

Culture, Collaboration, and Representation

Beyond the technical training, what stood out to me most at Mewburn Ellis was the work culture — something you cannot fully appreciate until you are immersed in it. From day one, it was clear that the firm places a strong emphasis on fostering a supportive and collaborative network. There was no sense of hierarchy getting in the way of learning; everyone, from Trainees to Partners, was approachable and genuinely invested in our development. It created an environment where questions were welcomed, discussions were encouraged, and growth was clearly a shared priority. What was especially refreshing — and frankly, motivating — was the number of women in senior, leadership, and partnership roles. In an industry that has historically skewed male, seeing so many women confidently leading technical discussions, shaping legal strategy, and mentoring others was both empowering and inspiring. Mewburn Ellis did not just talk about inclusivity — it was evident in their day-to-day operations, team dynamics, and who had a seat at the table. As someone early in my career, that kind of representation and openness left a lasting impression.