Why AI adoption stalls: the diagnosis most firms avoid

For all the discussion around legal AI, one assumption still persists: if firms choose the right tools, adoption will follow.

The reality is more complicated.

At LegalTechTalk's panel Why AI Adoption Stalls: The Diagnosis Most Firms Avoid, the conversation quickly moved beyond software, models and procurement. Instead, the discussion focused on something less visible, but arguably more important: how people change the way they work.

Across the session, speakers from legal innovation, knowledge management, technology leadership and legal engineering shared different experiences. Yet a common theme emerged.

AI adoption is about trust, not technology.

‍

AI adoption as a concept

AI adoption has become a vague concept. Firstly, AI spans across a plethora of technologies, that all require different skillsets to evaluate and leverage. For this purpose, we are talking about input-driven LLMs with agentic capabilites. Secondly, adoption implies a binary parameter. Either you have it or you don’t. In reality, adpotion is a journey with many layers.  

The end goal is clear. We want to leverage AI to assist legal professionals in their mission to uphold the rule of law. This clear end goal definitions leaves us with a couple of perspectives.  

Firstly, an adoption journey will almost never stop. As technology is growing, new ways to leverage it is growing accordingly. There’s no better time than the present to get started on your journey.

Secondly, this clear end goal requires steps on the way to the end goal to contribute to upholding the rule of law. In that sense, we need to acknowledge why legal professionals specifically bear the burden of this responsibility, and what processes have been set out to do so. Automating faulty processes while using AI can be considered adoption by some, although it is not contributing to the end goal. We see AI adoption as responsible inclusion of AI in existing processes placing legal professionals in the center, maintaining their role and responsibility, but with a greater arsenal of tools. In that sense, creating cognitive friction is a strength.

AI is changing the game rapidly.  

Automating faulty processes may contribute to the erosion of the rule of law, effectively countering the purpose of AI adoption.  

Responsible AI adoption, placing the legal professional in the center, will amplify its advantages.

‍

People adopt what they can see

One of the strongest messages from the panel was that adoption often begins with peer influence rather than formal training.

Several speakers reflected on the impact of senior lawyers demonstrating how they use AI in practice. Seeing a colleague apply a tool to real work creates a level of credibility that atraining programme usually can not replicate. Personal "eureka" moments and peer-led examples were repeatedly cited as catalysts for adoption.

However, these formal trainings vary in nature. Interactive trainings that are tailored to the users, can be a catalyst for creating personal “eureka” moments. When formal trainings are interactive, and focused around pedagogical principles and the art of conveying, these trainings can create a snowball effect.

In this sense, there are layers to the question. Personal eureka moments create traction. Good trainings create personal eureka moments.  

This matters because lawyers are naturally sceptical. And scepticism is far better than naivety.  

AI hallucinations are often considered one of the main risks concerning AI usage. However, the risk is not that AI hallucinates. The risk comes from using it without understanding its limitations or applying appropriate professional judgment. Our inherent scepticism as lawyers becomes a strength, as long as we are willing to demystify AI and trust our sceptic nature.  

Sceptics might start out arguing that AI is bad at the law, dumb and writes in a horrible style. However, these are all stances we effectively can disprove and move past quickly. These are not huge barriers to adoption. On the other hand, naivety can be.

The firms making progress are not asking lawyers to suspend their scepticism.

They are helping them apply it constructively.

‍

Training has changed

The discussion also challenged traditional ideas about training.

Rather than relying exclusively on structured courses, several speakers described approaches centred around experimentation, practical exercises and safe environments for learning. The goal is to help them become comfortable using it, not just holding a lecture.

This shift reflects a broader reality.

There is no definitive manual for AI.

The technology evolves quickly. New capabilities emerge constantly. Effective adoption therefore depends less on memorising instructions and more on developing confidence, curiosity and critical thinking.

The firms seeing the strongest results are creating opportunities for people to explore, test and learn in ways that are directly connected to their work.

‍

Focus on progress, not perfection

Perhaps the most practical lesson from the panel was a warning against waiting for perfect conditions.

Knowledge management matters. Data quality matters. Governance matters.

But several speakers cautioned against allowing those considerations to become reasons not to begin. Organisations can spend months debating readiness while missing opportunities to build understanding and momentum.

AI adoption is built layer by layer, at varying speed. It is not a linear process.

There are periods of enthusiasm, periods of resistance, and periods where progress appears to stall. According to one panellist, this is not evidence of failure. It is simply how adoption works. Every major technology shift follows a similar pattern.

Friction can and should not be eliminated, it just needs to be endured.

‍

What successful firms are doing differently

If there was one conclusion shared across the panel, it was that successful adoption requires intentionality.

The firms making progress are communicating consistently. They are creating opportunities for experimentation. They are investing in skills, not just software. They are identifying champions, building communities of practice and helping people understand where AI fits within existing workflows.

Most importantly, they recognise that people are protagonists.

AI can accelerate legal work, improve access to knowledge, support better execution and ultimately assist our mission of upholding the rule of law. But its value ultimately depends on the expertise, judgment and accountability of the professionals using it.

The firms that understand this are not simply using AI, calling it succesful adoption and deeming the job finished. They are creating cognitive friction, integrating the legal professional and AI side by side in the existing processes, and evolving alongside the technology.

‍