Re:Think 3 July, 2024

Winning the AI race

This blog was published to stimulate conversation ahead of Reform's roundtable with James on the role of AI in a thriving legal sector, kindly supported by the Solicitors Regulation Authority. 

The UK digital sector has largely missed out on the major technological revolutions of recent decades, from the creation of the internet to the rise of the smartphone and social media. That’s why none of the world’s 50 largest software companies (by revenue) are based here.

The AI revolution will reset the board.

Within the next decade, AI is going to completely change the way that most knowledge work is done. Given that the UK’s strengths are focused in knowledge industries which are prime candidates for AI disruption, it gives us the perfect vantage point to build AI applications for these industries, make up for missed opportunities, and boost flatlining productivity, which has barely grown since 2007.

While the developers of large foundation models, such as OpenAI and Anthropic, have attracted the most attention in the AI conversation so far, applications like biotech, finance, legal and professional services will soon produce a new generation of huge specialised AI companies. The UK can be home to these future winners that could well end up as big as Amazon, Google and Facebook, the winners of the last revolution, are today.

AI in practice: what does the legal sector of tomorrow look like?

• Today’s best large language models (LLMs) are very capable, when used in the right way. They can already score in the top decile of students at graduate-level legal exams. The next generation of LLMs, even more powerful, are already being trained, and the investment in GPU clusters to build the next generation after that is in development, at a cost of hundreds of billions of dollars. Expect to see a continuation of the shockingly rapid improvement in LLM capabilities, in turn bolstering the capabilities of start-ups that build their own AI on top of these foundational models in order to serve specialist markets such as the legal sector.
• Don’t expect to see fewer lawyers. As AI makes lawyers much more productive, demand will go up as good legal work will become faster, and more affordable. Just as the spreadsheet made accountants more valuable, rather than making them all redundant, legal AI will help meet unmet demand for legal services, and broaden access. The Legal Service Board recently found that around 3.6 million people who need legal services go without them each year in England and Wales.
• It’s the end of the billable hour. Paying an hourly rate used to make sense, but soon, tech-enabled lawyers might be 100 times more efficient than their old-fashioned competitors. An hourly rate disincentivises efficiency whereas billing for output encourages it, and incentivising efficiency is key to unlocking productivity gains.
• Humans will take on the role of commercial decision-maker and advisor, rather than be trained to be very good at carefully reading long documents. Much of a lawyer’s time today is spent on tedious document comprehension – drudgery which will soon be done by AI instead, leaving human experts to be strategic business partners.
• In-house legal teams will do more, using AI to perform research, to draft and negotiate contracts, write reports and perform due diligence, and to manage their commercial obligations. Industries with big legal bills today, like private equity, banking, insurance and pharmaceuticals, will be able to leverage technology to provide the capacity they use law firms for today. Combining AI with the commercial expertise of their in-house teams will enable them to move faster, save money, and outcompete those who don’t.
• Valuable business data, which is today locked away inside thousands of unread reports, memos and contracts, will be unleashed by letting AI read every page in seconds. This will boost the productivity of large corporations in particular. Security concerns will push these big enterprises towards partnerships with a small number of trusted AI providers whose access to that data will drive a snowball effect, further improving their capabilities.

2050 is just 26 years away. Google was founded 26 years ago, Amazon and Nvidia, a few years earlier, and Facebook just a few years later. These companies are all worth over a trillion dollars today. Nvidia is worth more than the whole FTSE100 combined.

Winning this race matters – it may matter more than everything else being done to grow the UK’s economy. And losing it will be costly. If our professional services industries don’t adapt to use AI quickly, they could disappear entirely. The transition to being AI-first will be uncomfortably fast, and it will be turbulent. But it is coming. The UK needs to drive this change rather than play catch up. That means an agile, pro-innovation approach to regulation, and doing everything we can to enable the UK’s technology start-ups to become the next global winners.