Highlights from UKTN's podcast featuring Robin AI CEO, Richard Robinson

AI will transform the way that lawyers work ‍

Richard Robinson, CEO and Founder of Robin AI, spoke to UKTN host, Jane Wakefield about the applications of generative AI in legal, Robin AI's journey, and the planned AI Summit later this year. He also provided some top tips for Founders thinking about fundraising and his views on the ongoing challenges in diversity in tech. We delve into some of the topics discussed in the podcast, which you can listen to in full here.

ChatGPT and hallucinations

Richard and Jane discussed the recent case that has made the headlines around the world as two lawyers in the US who had used ChatGPT to do case law research and make submissions to a court were fined $5000. The situation was a bit of a disaster: ChatGPT had hallucinated case law which the lawyers then used in submissions without verifying that they were real genuine cases.

This was an eye-opener for lawyers around the world about the risks of using generic tools like ChatGPT, which aren't tailored to the unique requirements of lawyers and legal work. Richard emphasised in the podcast the importance of finding the right partner when lawyers consider using generative AI, and the importance of identifying the correct use cases for it, where the lawyer can validate the outputs.

Richard detailed more information about Robin AI’s approach – and the early partnership with Anthropic. It’s important that lawyers understand that the AI is not there to replace them but to work alongside them and empower them to be able to work faster. Richard said, “it’s why it’s called a copilot.” It was also discussed how Anthropic’s approach was more suited to the requirements of Robin AI and the legal sector, as the company focussed on making their model “safe.” Which in the context of legal AI is all about ensuring that the model doesn’t try to answer questions that it doesn’t know the answer to.

Robin AI’s contract copilot provides detailed and specific outputs when answering questions about contracts. This means that if you ask a question of the contract copilot like, “what does confidential information mean?”, it will answer the question with reference to the contract that it is currently assisting with, and will be able to help guide the lawyer or the user to the relevant clause in the contract. This demonstrates the importance of using a contract copilot as a means to work faster alongside a lawyer, rather than replacing a lawyer during these processes.

Applications for large language models (LLMs) in legal

In the podcast, Richard describes the moment where he and his colleagues at Robin AI had begun to explore the capabilities of LLMs and had their “wow” moment. “We knew that we were at the beginning of a complete transformation in how legal work was done,” commented Richard. The team at Robin AI had been exploring the capabilities of LLMs to answer questions based on huge amounts of text, in this case, 150+page contracts. Richard remarked that the answer to a question might be buried in one sentence in over 100 pages of text, but that the models were able to answer the questions with 100% accuracy.

Robin AI’s team of AI and legal experts had been building AI products since 2019 and were fortunate to be early launch partners of Anthropic, which gave access to their LLM before it was more widely released. But even Richard and the team were impressed with how rapidly the transformation and evolution of AI developed, “we didn’t expect this to come this quickly.”

At the beginning of 2023, Robin AI launched the contract copilot, which leveraged LLMs to assist with a number of legal tasks such as answering questions about contracts, summarising information and translating complex legal language into simple phrases. The team have subsequently built a number of new features that harness the power of LLMs, such as the ability to use natural language to query entire contract repositories. This has enabled in-house legal teams across a range of industries to find information buried in their contracts in seconds. Developments with LLMs have also facilitated the evolution of AI suggested edits for contracts, including on third party paper, meaning that contract review can be accelerated by over 80% using Robin AI.

Robin AI’s 2023 GC Report, which was published in October, highlighted that whilst in-house legal teams were very interested in leveraging technology to support the long-term success of legal their departments, less than half of GCs had conducted any kind of formal assessment as to whether generative AI could support their teams. There’s still a lot of progress to be made to bridge the educational gap about the potential use cases and applications for large language models in legal. The Robin AI team have produced a guide to the applications of generative AI in legal to support this process.  

The decision to enable everyone to create free accounts with Robin AI

Richard and Jane discussed Robin AI’s contract copilot, and the decision that Richard made to enable everyone to create a free account with Robin AI. Jane highlighted that for many, “the law is seen as inaccessible.” Richard agreed, commenting that one of the reasons that he wanted to build Robin AI, was to democratise access to legal services. He noted that after qualifying as a lawyer at 24, his own parents couldn’t afford an hour of his time. Richard has spoken in more detail about his vision and mission for Robin AI, which you can watch here.

Robin AI’s contract copilot is used by individuals, organisations, and businesses around the world. Providing a free account that gives everyone access to the AI contract copilot not only helps to make the law more accessible, but it also helps lawyers to get more confident and comfortable with how they can boost their productivity and efficiency.

There’s another reason for providing free access to Robin AI’s accounts: helping lawyers to separate fact from fiction. As legal technology companies seek to try and jump on the excitement about generative AI and it’s transformative potential, there has been an increase in legal technology companies who are promising more than their products can deliver, or who are advertising products that don’t exist yet. There’s more information that you can read on Robin AI’s strict “show, don’t tell” policy.

AI safety

With the upcoming AI safety summit and the creation of the UK Government’s taskforce chaired by Ian Hogarth, an investor in Robin AI, Richard and Jane discussed the type of regulatory approaches that might be suitable in a fast-evolving landscape.Richard highlighted the importance of bringing the right expertise into the discussion, “the UK Government has done the right thing – they’ve appointed experts.” Richard pointed to the need for a balanced approach, “we want to be safe, but have enough space for innovation.”

Robin AI’s Co-Founder and CTO, James Clough commented on the upcoming AI safety summit for Startups, “Many people believe that AI safety regulations are a cynical ploy towards regulatory capture of the sector by Big Tech. Including earlier stage startups is one way to avoid that occurring, and that perception which may otherwise undermine support for important safety regulation.”

Diversity in Tech

Jane ended the podcast by asking Richard about diversity in the tech space. Richard remarked that this continues to be a “huge problem” and that change is slow, “we haven’t seen much progress, certainly since I started fundraising.” Innovation comes from all corners, and it’s important that business leaders continue to work to solve these challenges. Richard notes that he goes above and beyond to try to help other business leaders from underrepresented backgrounds, “I try to use the opportunity that I’ve been given to the maximum potential.”

Whilst Jane and Richard’s comments were focussed on the lack of Founders from underrepresented backgrounds, there remains an ongoing challenge with diversity in tech more generally. It’s a core part of Richard’s vision for Robin AI to “be one of the most diverse tech companies.” And Robin AI is making great progress, with 50% of board seats held by women, and 31% of the workforce represented by people of colour.