Regular readers of this website may have noticed that I have been rather quiet of late. That is not because there has been nothing happening in the world of costs. Quite the opposite. Costs law has retained its customary ability to generate new judgments, new arguments and, from time to time, entirely new problems which nobody had realised existed until somebody decided to litigate them.
The explanation is rather more prosaic. Most of my writing time over the last two months has been swallowed by a book. The book is Andrew and the Marvellous Analytical Engine: A Practical AI Primer for Lawyers. The second edition runs to nearly 100,000 words. This means that what originally started as an attempt to explain artificial intelligence to lawyers in reasonably straightforward terms became something considerably larger.
The central idea behind the book is a simple one. Artificial intelligence is going to change legal practice. Indeed, it already is. The interesting question is no longer whether lawyers should use AI, but how they should use it, what it can sensibly be trusted to do, and where its limitations lie.
There is a great deal of noise surrounding artificial intelligence. At one end of the spectrum are those who believe that it will shortly render most lawyers redundant. At the other are those who regard the whole subject as little more than another technological fashion which can safely be ignored until it goes away.
I do not think either view is particularly useful. AI is a tool. A remarkably powerful tool, but a tool nonetheless. Used properly, it can assist with research, analysis, drafting, document review, organisation and the innumerable small tasks which consume a lawyer’s working day. Used badly, it can produce confident nonsense at remarkable speed.
The book is therefore intended as a practical guide for lawyers who want to understand what this technology actually does and how it can be incorporated into legal practice without abandoning professional judgment along the way. It looks at the technology itself, the principal tools now available, practical applications in legal work and the difficulties which accompany them: accuracy, confidentiality, hallucinations, professional obligations and the continuing need for a human lawyer to understand, check and take responsibility for the work.
The second edition is now available to purchase through Amazon in Kindle, paperback and hardback editions.
Finishing the book also caused me to look again at this website. Costsbarrister.co.uk has existed for many years. Like a lot of professional websites, it had accumulated material gradually. Articles were added. Pages were altered. New sections appeared. Old ones remained. Nothing was particularly disastrous, but the whole thing had started to show its age.
So, once the book was finished, I turned to rebuilding it. The interesting part is how the rebuilding was done. I used ChatGPT extensively, not merely to suggest new wording, but to analyse the website as a whole. What was each page actually for? What information did a solicitor looking for costs counsel need? Was the important material obvious? Were cases being presented properly? Were articles easy to find? Did the site explain clearly what work I undertake and how I can be instructed?
This was coupled with what has rather unfortunately acquired the name “vibe coding”. Vibe coding does not mean asking an artificial intelligence to wave a magic wand over a website and hoping for the best. At least, it should not mean that.
In practice, the process was much more controlled. A problem was identified. A proposed solution was considered. ChatGPT was asked what was wrong with that solution. The proposal was revised. Then the change was implemented through AI-assisted WordPress tools and the actual page checked afterwards.
One of the lessons from rebuilding the site is that artificial intelligence can be a very capable implementer, but it is also an extremely literal one. If the instruction is poor, the result will usually be poor with impressive efficiency. The useful skill is therefore not simply telling AI what to do. It is deciding what ought to be done in the first place.
The result is a substantially rebuilt costsbarrister.co.uk: clearer navigation, better presentation of cases, improved articles, more obvious routes for solicitors wishing to instruct me and, I hope, a website which is considerably easier to use.
It also means that I can return to the original purpose of the site. Writing about costs. There is no shortage of material. Detailed assessment, costs budgeting, Part 36, QOCS, solicitor and client costs, litigation funding, security for costs and the many other corners of this jurisdiction continue to produce issues worth examining.
So the recent silence should now come to an end. The intention is that articles will appear here much more regularly. Some will deal with important new decisions. Others will revisit points of costs law which deserve rather more attention than they usually receive. And some will look at the wider changes taking place in litigation, funding and legal practice.