The Human Element
Why AI makes regulated practitioners more valuable, not less
This essay sits primarily under the How do we do AI? question of the practice’s methodology: why AI tends to make regulated practitioners more valuable, not less, when adopted with care.
The fear
Many people in your firm, perhaps including you, are worried that AI will make them irrelevant.
Junior solicitors worry the work they were going to learn from will be done by a model. Junior accountants and graduates wonder what they will do that a chatbot cannot. Senior practitioners watch the headlines about disruption and run the numbers on how many years they have left.
The fear is understandable. It is also mostly wrong: people consistently misjudge how powerful technologies change the world.
The radiologist fallacy
In 2016, prominent AI researchers predicted that AI would replace radiologists within five years. Radiology was supposedly the canary in the coal mine: a profession built on pattern recognition that AI could do faster and cheaper.
Today, demand for radiologists is at an all-time high.
What happened? AI tools made medical imaging cheaper and faster. Instead of replacing radiologists, this led to an explosion in the volume of scans ordered, which created more demand for human expertise in complex diagnosis, treatment planning, and patient communication.
The prediction got the technology right but the economics wrong.
Jevons paradox: efficiency creates more work, not less
In 1865, economist William Stanley Jevons observed something counterintuitive: when James Watt’s steam engine made coal use more efficient, total consumption of coal did not decrease. It increased dramatically. More efficient use of a resource increased demand for it.
This pattern, Jevons paradox, has repeated throughout history:
| Technology | Expected effect | Actual effect |
|---|---|---|
| Containerised shipping | Fewer dock workers | Global trade exploded; massive new industries in logistics |
| Cloud computing | Fewer IT jobs | Server admins became DevOps engineers managing at impossible scale |
| ATMs | Fewer bank tellers | Cheaper branches meant more branches; teller numbers rose |
| Spreadsheets | Fewer accountants | Demand for financial analysis skyrocketed |
| Desktop publishing | Fewer designers | The volume of designed content exploded |
The pattern is consistent: when the cost of doing work goes down, demand for it goes up. Usually there is far more latent demand than anyone realised.
What this means for professional services
As AI makes drafting, research, analysis, and summarisation cheaper, you will not necessarily have fewer practitioners. You will likely see a massive increase in the volume of regulated work: more matters, more advice, more complex questions clients felt previously priced out of asking, and a corresponding need for humans who can direct, evaluate, and make judgement calls about quality.
The lower the marginal cost of a draft, the more drafts get produced, and the more important it becomes to have someone who can spot which ones are wrong.
Roles do not disappear. They get refactored. The shift is from manual execution to supervision, from routine tasks to complex judgement, from doing the work to defining what good work looks like.
What clients actually buy from a regulated firm
If AI can produce a passable first draft of a contract, a tax position, a clinical letter, or a Statement of Advice, why would anyone keep paying a firm?
For the same reason they kept paying universities after MOOCs, accountants after spreadsheets, and lawyers after Westlaw. Content was never the main value proposition.
What clients actually buy:
Curation and structure. Someone has to decide what the question is, what is worth doing, in what order, at what depth. AI can generate a draft, but deciding which draft this client needs, and which they should not be sent, requires human judgement about the matter, the regulator, and the relationship.
Accountability. Your name is on the advice. Your registration is on the line. If the answer is wrong, AI does not get sued. The client buys someone who is willing to stand behind the work.
Discretion and confidentiality. Clients tell their lawyer, GP, accountant, or financial adviser things they would not put into a chatbot. That trust is not something AI can manufacture. Maintaining it, and being seen to maintain it, is part of what they pay for.
Judgement under uncertainty. Real matters have ambiguous facts, contested law, irrational counterparties, and political dynamics. AI optimises against a defined objective. Practitioners decide which objective is even worth optimising for.
Adaptability. A clinician notices the question behind the question. A solicitor reads what the client did not say. An accountant catches what is missing from the records. These adaptive, contextual responses remain difficult for AI to replicate with the same depth.
Every technology that promised to replace professional services has been absorbed by them: books, photocopiers, fax, email, the internet, cloud accounting, document automation. Firms adopted them all as tools while continuing to provide the human elements technology could not replicate.
AI will follow the same pattern. It will not replace your firm. It will be absorbed, and the firms that use AI well will provide better service than those that do not, because their practitioners will be free to do the work AI cannot.
The human premium
So what are the human capabilities that become more valuable as AI becomes more capable?
The shortest answer is agency: the ability to steer, choose, and stand behind the work. The capabilities below are all expressions of it. Each of them is something AI can support but cannot replace, because each requires a person who is willing to be the source of the decision rather than the editor of the model’s first draft.
Judgement
AI can generate options, analyse data, and draft a position. Practitioners decide which options are worth pursuing, what the data means in this client’s context, and whether the position is sound, ethical, and aligned with the client’s actual interests.
Judgement requires experience, context, and values, none of which can be fully encoded in a model.
Ethical reasoning
AI can identify patterns and optimise for defined objectives. Defining which objectives are worth optimising for is an inherently human question. Should the firm optimise for efficiency or thoroughness? Speed or accuracy? The client’s stated wish or the client’s actual interest?
These are professional questions, not technical ones: the questions that show up in your code of ethics, in APP, in APES 110, in AHPRA’s expectations, in your conduct rules. They become more important, not less, as AI makes it easier to optimise at scale.
Relationship and trust
Clients trust a practitioner who knows their history. Staff trust a partner who understands their constraints. Counterparties trust someone who listens before advising.
Trust is built through human interaction: through vulnerability, consistency, and demonstrated care. AI can simulate empathy. Humans can be empathic. The difference matters, and clients can tell.
Contextual understanding
AI knows a lot about the world in general. You know a lot about your world in particular: your clients, your firm’s appetite, your local regulator’s preferences, your community, your constraints. This contextual knowledge is irreplaceable, and it becomes more valuable when combined with AI’s breadth.
Accept mediocrity, or elevate practice?
Option A: Accept mediocrity at scale. Automate everything and replace judgement with AI confidence. Send more letters, faster, without proper checks. Build a team that can use tools but cannot think critically. The firm looks efficient until a complaint arrives and the insurance does not cover it.
Option B: Elevate what makes the firm worth hiring. Use AI for the routine work so people can advise, mentor and treat patients. The work gets better, and there is more of it.
Option A is easier. Option B requires intentional effort. Most technologies used for amplification rather than replacement have produced better outcomes and more durable businesses.
What this means for you
If you are a principal or partner
You are not being replaced. You are being freed from the least valuable parts of your work (repetitive drafting, formatting, low-level summarising, administrative tasks) so you can focus on the most valuable parts (mentoring, reviewing, advising, building the relationships that bring matters in the door).
The principals who thrive will be those who understand AI well enough to use it as a thinking partner, and who double down on the human elements that make their firms worth hiring.
If you are a staff member
Your institutional knowledge, relationship skills, and contextual judgement become more valuable, not less. AI can draft emails and process forms. It cannot navigate office politics, exercise discretion in sensitive situations, or build the trust that makes a firm function.
The skills worth investing in are the ones AI makes more important, not less: critical thinking, ethical reasoning, communication, collaboration, and the ability to evaluate whether AI output is actually good.
If you are early in practice
The work that used to be your training ground (the first-draft letter, the file review, the basic research) will increasingly be done by AI. That sounds threatening. It is also an opportunity. You will have access to more matters, more variety, more senior conversations earlier than the people who trained you did.
The catch is that you have to build the underlying capability anyway. The graduate who skips the foundation because AI produced a draft has not saved time. They have skipped the part where they would have become useful. Use AI to extend yourself, not to replace yourself.
The pattern across history
| Era | Technology | Fear | Reality |
|---|---|---|---|
| 1440s | Printing press | Scholars would become irrelevant | Literacy and scholarship exploded |
| 1800s | Industrial machinery | Craftspeople would be replaced | New skilled trades emerged |
| 1980s | Personal computers | Office workers would disappear | Knowledge work expanded massively |
| 1990s | Westlaw / online research | Lawyers and librarians would be redundant | The market for legal expertise grew |
| 2000s | Cloud accounting | Bookkeepers and accountants would be replaced | Demand for advisory work expanded |
| 2020s | Generative AI | Professional expertise will be devalued | You are here |
The pattern is consistent: transformative technology amplifies human capability rather than replacing it, when humans actively shape how it is used.
The “when” matters. Technology does not automatically improve things. It takes people who understand both the technology and the professional context to ensure it is used for amplification rather than replacement.
The bottom line
AI is going to change what your firm does, not replace it: what you bill for, what you train juniors in, what counts as good work. The routine parts will be automated. The parts that require judgement, ethics, relationships, and contextual understanding will become your primary contribution, and they will be what clients keep paying for.
The question is whether you will develop the skills to work with AI in ways that make you more valuable.
The human element is the part AI makes essential, and at the centre of it is human agency.
About this work
This is the argument The Human Edge develops. borck.consulting helps regulated Australian small firms apply that judgement in practice, without losing the human element clients are actually paying for. Engagements begin with the AI Readiness Diagnostic.
Further reading
- Autor, D. H. (2015). Why are there still so many jobs? The history and future of workplace automation. Journal of Economic Perspectives, 29(3), 3–30. https://doi.org/10.1257/jep.29.3.3
- Acemoglu, D., & Restrepo, P. (2019). Automation and new tasks: How technology displaces and reinstates labor. Journal of Economic Perspectives, 33(2), 3–30. https://doi.org/10.1257/jep.33.2.3
- Deming, D. J. (2017). The growing importance of social skills in the labor market. The Quarterly Journal of Economics, 132(4), 1593–1640. https://doi.org/10.1093/qje/qjx022
- Wilson, H. J., & Daugherty, P. R. (2018). Collaborative intelligence: Humans and AI are joining forces. Harvard Business Review, 96(4), 114–123.
- Agrawal, A., Gans, J., & Goldfarb, A. (2018). Prediction machines: The simple economics of artificial intelligence. Harvard Business Review Press.
- Brynjolfsson, E., Mitchell, T., & Rock, D. (2018). What can machines learn, and what does it mean for occupations and the economy? AEA Papers and Proceedings, 108, 43–47. https://doi.org/10.1257/pandp.20181019