LONDON — Many legal practitioners, hearing the current hype around the application of artificial intelligence (AI) to legal problems, assume this is all new stuff. But data scientists specializing in those issues have met biennially at the International Conference on Artificial Intelligence in Law (ICAIL) since 1987.
And that’s a symptom of a problem: many in the field have a sense that the research and new techniques coming out of an academic conference like ICAIL are not reaching and influencing the legal profession. Similarly, many practitioners who are otherwise interested in and aware of some of the possibilities of AI feel that sometimes the academic researchers aren’t dialed in to real-world issues that the new techniques could address.
To fill those gaps, the ICAIL conference in London this year included a well-attended pre-conference workshop on AI and Legal Practice on Monday.
The day was organized by current International Association of AI and Law President Katie Atkinson, local host Jerome Keppens of King’s College, and Nick West, Chief Strategy Officer of Mishcon de Reya.
On the theory that the best defense against market hype is a little well-grounded understanding, the workshop was an all-day mix of presentations and discussions intended to bridge that gap between theory and practice. About 95 registrants attended, fairly evenly divided between law firm representatives, academics in data science fields and from law schools, and legal technology providers.
David Halliwell, Director of Knowledge and Innovation Delivery for Pinsent Masons, opened the event with a pretty blunt assessment: we are at the peak of a hype cycle on AI in the law now. Many of the attention-grabbing announcements by law firms of their latest AI investments are really just hype. But behind the hype, real progress is actually underway, Halliwell said, pointing out that AI isn’t just one technology, and that it has more than one application in legal work. He identified five areas where AI is having some impact already: Automation of legal reasoning; large-scale document review; Financial analytics for better business decisions; legal research; and prediction of legal outcomes.
So much of AI is shrouded in mystery to people outside the data science field, but their straightforward explanations of concepts… brought it all down to earth.
Many attendees thought the heart of the workshop was two tutorials, in which Isabel Sassoon and Qi Hao, data scientists from King’s College, walked through some terminology and applications for Machine Learning and Natural Language Processing.
It was an important presentation. If lawyers are to become more comfortable with AI, one good place to start is just acquiring a basic understanding of terminology and the various techniques of AI.
These sessions gave the lawyers in the audience a peek inside the black box. So much of AI is shrouded in mystery to people outside the data science field, but their straightforward explanations of concepts such as supervised vs. unsupervised learning, segmentation, clustering, text summarization, information extraction, etc. brought it all down to earth. Introductions to basic AI concepts like this, if they become more common in law schools and in ongoing professional development for lawyers, will go a long way to de-mystify the discussion, drive out some of the hype, and perhaps arm lawyers with the questions they need to ask in order to distinguish AI hype from useful applications.
The day also included some hands-on demonstrations from legal tech companies such as Brainspace, Neota Logic and Kira, all of which currently have solutions in the market that leverage various technologies on the AI spectrum. It wrapped up with a lively discussion among all participants about possible future steps.
In her Presidential Address to the full conference the following day, Atkinson summarized some of those ideas and imperatives:
- Developing a shared language for lawyers and researchers;
- “Fixing the piping before thinking about the magic”;
- Resolving IP issues when working through partnerships;
- Addressing the lack of shared datasets to work on;
- Addressing the lack of open source software; and
- Needing to focus on real legal tasks.
One thread that came out of this discussion was the ever-present tension between using more simple and proven technologies to automate away some of the more obvious inefficiencies in legal practice, on the one hand; and reaching for the more transformational changes to legal practice that AI can bring, on the other. There is a lot of low-hanging fruit that can be addressed with well-established technologies, but those with a longer time horizon don’t want to ease up on developing the cutting edge.
Atkinson clearly sees better alignment with lawyers and with commercial legal tech providers as key to the success of the AI and Law community’s future. Indeed, recent steps in that direction seem to be working — for example, attendance is up at conferences such as this one. Academic communities like these are operating in a dynamic environment, with law firms, legal and commercial media, legal tech events and legal startup communities all rushing to jump on the AI bandwagon. The academics don’t want to be left behind, and the profession should want them to be at the table as well.
As usual, the poets have the last word. Much of the resistance to AI is about a failure to break down self-imposed ideas about how legal services should be organized.
Milos Kresojevic of Freshfields Bruckhaus Deringer cited the 13th century Persian poet Rumi:
“Your task is not to look for love, but merely to seek and find all the barriers within yourself that you have built against it.”
The same, Kresojevic argued, can be said about AI. It makes no sense to seek AI as an end in itself; it’s only a tool that we can use to re-think the many ways legal services can be delivered. The focus should be on the practice and the client’s needs, not on the tools.