Mind The Gap : AI Adoption, Equity And Thinking Skills At Stake

When I describe myself as an arti cial intelligence (AI) expert in the legal technology space, I inch. It’s not that I doubt my abilities, or worry that my team and I aren’t delivering an excellent end-to-end service: from sourcing and developing GenAI tools and piloting and implementing them to transforming knowledge management (KM) into a high impact, revenue-generating function and championing the adoption of new technologies and ways of working. We are doing that, and more. So, why am I uncomfortable calling myself an expert? Why am I seeing a troubling skew in the take-up and use of AI tools across race and gender lines?

IT’S UP TO EACH OF US, INDIVIDUALLY, NOT TO ALLOW OUR OWN UNDERSTANDING AND SKILLS TO BE SUBVERTED BY THE EASY ACCESS THESE GENAI TOOLS PROVIDE.

Skill, insight, ability and hard work have no gender, race or age. Self-belief, however, might. This is once again showing up in how AI is being adopted in practice.

Some of my colleagues recently published an alert about employment equity considerations when deploying GenAI. Around the same time, the KM team was developing data-driven reports on GenAI’s usage and impact in our rm. Out of curiosity (not expecting to see any gender- or race-based skew in the numbers), I asked the team to include demographic details alongside the other data. Perhaps it shouldn’t have surprised me, but I was genuinely taken aback to see that our user numbers didn’t match our demographic distribution.

ENCOURAGING EQUITABLE AI ADOPTION

This was an exceptionally useful exercise, and we’re now taking active steps to encourage adoption and effective usage across all relevant groups within the rm. However, it held up a mirror to our team (the vast majority of whom are female, with signi cant diverse representation): why was this happening, and what other unintended behaviours should we be watching for?

Our approach has been to make the data visible and open the discussion within the rm. By asking our colleagues why there is a discrepancy in user demographics, and why they are personally choosing to engage or not engage with the technology, we are holding that same mirror up to them, encouraging self-reection on what might be holding some of us back. This allows our colleagues to acknowledge a self-belief issue, should that be at play, and for us to initiate steps to support them in engaging with the discomfort of change.

Over time, we will focus on creating spaces where people can experiment without pressure to be procient. We are also actively modelling female expertise in this area, and celebrating progress, not just mastery. We don’t believe there are systemic barriers to equitable adoption of the tools, but it bears thinking whether implicit or indirect barriers to use are at play.

Beyond keeping an eye on equitable access to tools promising massive ef ciency gains for lawyers, another side of GenAI implementation keeps me awake at night: how do we guard against cognitive decline in experienced users while encouraging the development of critical thinking skills in everyone, especially those just entering the profession?

These tools are addictive. It’s all too easy to let our bias towards the results they produce lull us into a misguided trust in the outcomes. We hear the same fairly obvious statements repeated constantly: verify all content, use your own critical thinking skills, don’t rely on the tools for critical evaluation, and remain responsible for everything you generate. All perfectly true, and perfectly meaningless, at the same time. The reality is that it’s virtually impossible to enforce responsible AI use while safeguarding your own cognitive development. Ultimately, it’s up to each of us, individually, not to allow our own understanding and skills to be subverted by the easy access these GenAI tools provide.

Both my daughters are squarely within the generation that has grown up with instant and constant access to smart devices. Their thumb-typing skills are extraordinary – faster than I can manage with all ten digits. They’re used to engaging with their peers through an electronic lens. Even so, my eldest recently mentioned it has become increasingly common for her friends to ask their preferred GenAI model for appropriate responses to even the most mundane interpersonal conversations. Why? Why do we think platitudes from a large language model are more impressive or safer than our own words in everyday conversations? Won’t the consequence be that we’ll increasingly lose the ability to conceive and articulate ideas, from the mundane and everyday to the complex and advanced?

For those of us responsible for training and developing the next generation, this is the biggest challenge (and greatest opportunity) of our lifetimes. I wish Socrates was still around to help us conceive the perfect way to teach in an AI world. Or maybe he already has … pregnant pause.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.