AI isn't the hard part. Changing ourselves is.
- David Langdon
- 2 days ago
- 9 min read
AI Isn't the Hard Part. Changing Ourselves Is.
I recently spent a morning in a room full of lawyers, legal educators, technologists, consultants and legal industry leaders discussing artificial intelligence. Like many conversations on the topic, I expected much of the discussion to centre on prompts, large language models, regulation, hallucinations and the growing range of AI products entering the legal market.
Instead, we spent remarkably little time talking about the technology itself and a great deal of time talking about people.
That struck me because, after more than thirty years working in legal technology, I have witnessed almost every major wave of innovation that promised to transform the profession. Document management, knowledge management, workflow, cloud computing, mobility and automation have all changed the way legal work is performed.
Yet each has encountered the same constraint. The technology was rarely the hardest part. The real challenges were organisational:
Leadership Making clear decisions about why change matters and what the organisation is trying to achieve.
Culture Creating an environment in which people feel able to question established practices and try something different
Processes Understanding how work is actually performed before attempting to automate it.
Incentives Ensuring people are rewarded for adopting better ways of working rather than simply maintaining familiar ones.
Willingness to change Helping people move beyond knowing that change is required to behaving differently in practice.
AI feels different, not because those challenges have disappeared, but because the pace and scale of change are unlike anything we have experienced before. The technology is evolving rapidly, expectations are shifting just as quickly, and organisations are being asked to rethink long-established ways of working in months rather than years.
As the discussion unfolded, it became clear that AI was really just the starting point. The conversation quickly expanded into education, employment, governance, workplace design, client expectations and the nature of professional judgement.
These are not technology questions. They are questions about how the legal profession chooses to evolve.
The roundtable was ostensibly about artificial intelligence. By the end of the morning, it felt much more like a discussion about the future of the legal profession.
A Change Bigger Than Technology
For decades, law firms have worked to align technology with business strategy. AI has not removed that challenge. If anything, it has made it more urgent.
One participant reflected on graduates entering the profession wanting to solve complex problems and become trusted advisers, only to discover that much of their early career is consumed by repetitive research, document review and administration.
AI has the potential to reduce that burden and allow junior lawyers to contribute to more meaningful work much earlier in their careers.
That raises an interesting question.
If AI saves a lawyer five hours, what should happen to those five hours?
They could be invested in:
Mentoring and developing junior colleagues
Building deeper expertise
Undertaking more thoughtful analysis
Strengthening client relationships
Improving the quality and accessibility of legal services.
Equally, they could become five additional billable hours.
Efficiency has an extraordinary ability to create new expectations rather than simply creating more time.
That is not a technology decision. It is a leadership decision.
One of the most thought-provoking ideas raised during the discussion was that AI should not be viewed simply as another tool. A tool remains dormant until someone chooses to use it.
AI is becoming something much broader: a medium through which information is found, knowledge is created, work is performed and decisions are supported.
That distinction changes the conversation. Rather than asking only how lawyers should use AI, we also need to ask how AI will change the environment in which lawyers practise.
It is already reshaping:
Client expectations
Legal education
Supervision and professional development
Knowledge management
The economics of legal services.
In that sense, AI is not simply changing how legal work is performed. It is changing the profession itself.
Will AI Replace Lawyers?
No discussion about AI lasts very long before someone asks whether it will replace lawyers. It is an understandable question, but as the morning went on, I became less convinced it was the right one.
A more useful question might be:
What will clients and society value most in lawyers when AI can perform more of today’s routine legal work?
Some work will undoubtedly change. Research, summarisation, document comparison, drafting and information retrieval are already being transformed by AI. It would be difficult to argue otherwise.
The legal profession has adapted to technological change many times before, although it has occasionally behaved as though precedent might offer some protection.
What felt different in this discussion was that very few people believed the future was simply about doing the same work faster. Instead, the conversation turned to what lawyers should be doing with the time and capacity AI creates.
History suggests technology changes employment more often than it simply removes it. Some roles contract, others evolve, and entirely new areas of expertise emerge in response to opportunities and risks that did not previously exist.
Few firms once employed specialists in cybersecurity, privacy, legal operations, digital accessibility or user experience. These disciplines developed because technology changed the way legal services were delivered. AI is likely to follow a similar path.
As AI becomes capable of generating almost unlimited amounts of information, the ability to produce content becomes less distinctive.
The ability to do something useful with that content becomes far more valuable:
Evaluate it Determine whether the information is accurate, relevant and appropriate.
Challenge it Question confident answers rather than accepting them at face value.
Govern it Establish clear standards for how AI should and should not be used.
Apply judgement Interpret information within the legal, commercial and human context of a particular situation.
That shift suggests growing importance for capabilities that extend well beyond technology. Workplace experience, information governance, knowledge curation, AI assurance, learning and capability may become just as important as selecting the right AI platform.
These disciplines exist because organisations still need people who can distinguish authoritative knowledge from plausible-sounding output, establish appropriate governance and help others use new technology responsibly.
These roles will not emerge because AI has failed.
They will emerge because it has succeeded.
Knowledge is becoming abundant.
Judgement remains scarce.
The Human Element is Not Yet Replaceable
One theme surfaced repeatedly throughout the discussion, even when nobody said it explicitly. AI can perform many impressive tasks, but it still operates without the human context in which legal advice is given and received.
A client rarely arrives with only a legal question. They may also bring uncertainty, frustration, grief, commercial pressure or conflict.
The technical answer matters, but so does the ability to:
Interpret a complex situation
Understand what remains unsaid
Recognise the emotional and commercial context
Build confidence that someone genuinely understands the problem.
Those qualities are difficult to measure, yet they often determine the value of professional advice.
AI can recognise patterns associated with emotion and communication, but recognising a pattern is not the same as understanding the experience behind it.
The same applies inside a law firm. AI cannot fully appreciate the history behind a difficult client relationship, the dynamics of a partnership or the reason a process has survived years of attempted improvements.
It may identify inefficiencies, but it cannot always understand the organisational, cultural or political context that created them.
As the discussion progressed, I found myself thinking less about what AI can do and more about what continues to distinguish good professionals:
Sound judgement in uncertain situations.
Curiosity to question an answer that appears convincing.
The confidence to say, “I’m not sure that’s right.”
Empathy when clients need reassurance rather than information.
The ability to earn trust over time.
These qualities have always mattered, but AI may make them more valuable rather than less.
As technology becomes increasingly capable of generating answers, the human ability to ask better questions, interpret complexity and exercise judgement may become the profession’s greatest differentiator.
Perhaps that is the real opportunity. AI may take on more of the routine work, allowing lawyers to spend more time doing the work that only people can do.
AI as an Accessibility Equaliser
My own perspective on AI is shaped by a very personal experience. As someone who is legally blind, I rely on technology every day in ways I never imagined even a few years ago.
AI reads and summarises lengthy documents, helps me navigate complex information, critiques ideas, explains unfamiliar concepts and allows me to work with large volumes of content far more efficiently than traditional assistive technology ever could.
For me, AI has become much more than a productivity tool.
It has become an equaliser.
It has the potential to reduce barriers created by inaccessible documents, visually dense information and systems that were never designed with everyone in mind.
Used thoughtfully, AI can make legal workplaces, legal education and legal services significantly more inclusive.
At the same time, I left the discussion thinking about a contradiction.
The same technology helping remove barriers is also making it easier to create new ones.
AI-assisted software development is changing how websites and applications are built. Today, someone with limited technical experience can describe an idea in plain English and generate a polished website or application within hours.
That is an extraordinary step forward for innovation and creativity.
But polished does not necessarily mean accessible.
If accessibility is not part of the prompt, it is unlikely to be part of the result. A website may look modern while still presenting significant barriers for someone using a keyboard, screen reader or other assistive technology.
Common problems remain surprisingly easy to generate:
Poor heading structures
Inaccessible forms
Weak colour contrast
Missing alternative text
Unclear keyboard focus
Controls without meaningful labels.
AI is not creating inaccessible experiences because it lacks capability.
It is creating exactly what people ask it to create.
As AI lowers the barriers to software development, accessibility becomes even more important. It can no longer be treated as a compliance exercise completed at the end of a project.
It needs to become part of the conversation before the first prompt is ever written.
If AI is going to make technology more accessible, we need to ensure it is not simultaneously making the digital world less inclusive.
The Organisational Challenge
One of the themes that emerged repeatedly throughout the morning was that organisations often focus on technology before they focus on themselves.
Technology has always been the visible part of transformation. The harder work happens underneath.
Workplace experience is shaped by people, processes, information and technology working together. AI strengthens one part of that system while exposing weaknesses in the others.
Poorly designed processes become obvious when firms attempt to automate them.
Weak information governance becomes more risky when AI can retrieve information instantly.
Inconsistent ways of working become difficult to ignore when organisations try to scale them.
Accessibility gaps become more visible when new digital experiences are created at unprecedented speed.
One of the lighter moments during the discussion came when someone proudly demonstrated an AI application they had built, only for an experienced software engineer to reply:
“It’s great. Just don’t use it in a business.”
The room laughed because everyone understood the point.
AI has given many professionals capabilities they did not previously possess. It has not automatically given them expertise in software engineering, cybersecurity, governance, information architecture or organisational risk.
That distinction matters.
Many organisations will successfully deploy AI tools. Fewer will invest the same effort in improving the foundations those tools depend upon.
Adding AI to fragmented processes, unreliable information or unclear governance does not produce transformation.
It simply allows old problems to move faster.
The Blind Spots AI Reveals
As the discussion drew to a close, I found myself thinking less about what AI creates and more about what it reveals.
Perhaps AI’s greatest contribution will not be the content it generates.
Perhaps it will be the blind spots it exposes.
Duplicated work Tasks that everyone assumed were unavoidable because nobody had stepped back to question them.
Information that nobody can find Knowledge scattered across systems, inboxes, folders and individual memories.
Inconsistent processes Ways of working that have quietly evolved across teams, offices and practice groups.
Inaccessible systems Barriers that have excluded people without anyone deliberately intending them to.
Outdated governance Controls designed for a world in which information moved slowly, rather than one in which it can be generated in seconds.
None of these issues are new.
AI is simply making them more difficult to ignore.
That observation resonated strongly with me because it reflects something I have seen repeatedly throughout my career. Organisations often believe they have a technology problem when they actually have a workplace experience problem.
The technology simply brings it into sharper focus.
The firms that gain the greatest value from AI are therefore unlikely to be those deploying the largest number of tools.
They are more likely to be the organisations that understand:
How work is performed
How information flows
Where people experience friction
Which barriers are preventing adoption
Which improvements genuinely matter.
Technology can accelerate change.
It cannot replace thoughtful organisational design.
Changing Ourselves
As I sat on the train on my way home after the roundtable, I realised I had spent remarkably little time thinking about artificial intelligence itself.
Instead, I found myself reflecting on:
Trust
Judgement
Curiosity
Leadership
Education
Accessibility
Culture.
The conversation had begun with AI, but it had gradually become a discussion about people and the choices organisations will need to make as AI becomes part of everyday professional life.
How do we help junior lawyers develop judgement when AI can perform much of the work through which previous generations learned?
How do we encourage critical thinking when plausible answers are always only seconds away?
How do we govern AI without discouraging experimentation and innovation?
How do we use the capacity AI creates to improve legal services rather than simply expecting people to do more?
None of these questions can be answered by purchasing another software licence.
They require organisations to think carefully about leadership, learning, inclusion and the kind of profession they want to build.
The future of legal practice will undoubtedly be influenced by artificial intelligence.
But I suspect it will be shaped just as much by human intelligence:
By our willingness to question established ways of working
By our willingness to redesign workplaces around people rather than technology
By our willingness to invest in judgement as much as efficiency
By our willingness to use new capacity thoughtfully rather than simply filling it with more work.
History suggests technology has rarely been the hardest part of transformation.
Changing ourselves has always been the greater challenge.
Perhaps that was the most valuable insight I took away from the discussion.
