It’s hard to escape the noise about Generative Artificial Intelligence (GenAI) shaking up the legal world. The type of artificial intelligence that can generate new content, such as text, images and videos based on a user’s set of instructions (a prompt). Originally, there was talk about GenAI replacing lawyers, which is understandable given GenAI’s ability to recognise, interpret and generate text using advanced computing models that have been trained on large data sets (known as Large Language Models).
That messaging has since been softened to lawyers that use GenAI will replace lawyers that don’t (we can all take a deep sigh of relief). It got me thinking, what does this statement actually mean for lawyers, given many I speak to are taking a wait and see approach when it comes to GenAI? Do we all need to become experts at using GenAI, which currently means being able to expertly instruct (prompt) these tools? Will non-lawyers who can craft effective prompts based on their understanding of how these GenAI models interpret language (known as prompt engineers), replace lawyers that use GenAI?
In short, I don’t think every lawyer needs to become a specialist prompt engineer, as that assumes that the in-demand prompt engineering jobs of today will still be needed as GenAI technology evolves. GenAI tools are becoming increasingly more intuitive and user-friendly. The emphasis on the need for complex prompting is likely to diminish as GenAI’s natural language processing capabilities improve. Future GenAI models will be more adept at understanding nuanced instructions and yes, your legal jargon.
The companies developing these tools have a vested interest in the widespread adoption of GenAI as part of our day-to-day working life, so that they can recoup the high GenAI computing and power costs. To achieve this, they are actively working to make these tools more accessible to the average user by removing the prompting barriers (they don’t want you to have to employ an expert operator). Features like suggested prompts based on context, and automatic refinement of poorly worded prompts, are now built into GenAI tools. As these tools continue to evolve, the technical barrier to using GenAI tools effectively, will continue to drop. You can see where things are heading.
My personal experience with Gemini (Google’s GenAI tool) in a work environment has been positive - it helps with the blank page problem and creates efficiencies. When appropriate, I’ve been using Gemini’s suggested prompts as a starting point. I build out that suggested prompt based on prompting best practices and then progressively refine it until I achieve my desired output. The suggestions are usually relevant as the prompts take context from the application that I’m using at the time (such as email) and/or the content of the document I’m working in, or have referenced.
So where does this leave us…
I don’t think lawyers need to be GenAI super users today, but they need to be actively and regularly using GenAI tools with a view to it being part of their day-to-day working life tomorrow. We can’t sit on the sidelines - we need to experiment. Lawyers should continue to develop their analytical thinking skills as these help to build effective prompts. Being able to identify the relevant legal ‘ask’ (the problem and required output) and break down that complex ‘ask’ for a GenAI tool using natural language (problem decomposition), are the prompt adjacent skills that will endure and enable lawyers to be champions of tomorrow’s GenAI tools.
Ngā mihi nui ki a Mitch Julian for sharing his experience with our community.
Check out Mitch's website profile here and LinkedIn.