You may remember the ChatGPT lawyer case – the 30-year veteran who filed a brief citing six non-existent court cases, then got sanctioned $5,000 when opposing counsel couldn’t find them.
But here’s what I think we missed: this wasn’t really about a lawyer making a mistake. It was about something happening across every industry that we don’t have a name for yet.
I call it “borrowed competence.”
The lawyer spent 20 minutes getting work that looked like hours of research. He had a professional brief without doing the underlying work. The competence belonged to the AI – he was just using it on credit.
The thing about borrowed money is that eventually somebody calls in the loan.
I’m seeing this everywhere now, people using AI to draft consumer affairs complaints they can’t argue in mediation, bug bounty hunters submitting AI-generated reports for vulnerabilities they don’t understand, and business owners ditching “costly” legal advice in favour of AI.
Most of the time, the work looks competent.
The problem is that the user hasn’t built the underlying capability and has no way of knowing when it’s wrong.
Three years into the AI era, regional Australia faces a choice: Are we using AI to build capability our people actually own?
Or are we just renting competence we’ll eventually have to repay? I think this question matters more than we realise.
And I think the answer could give us a genuine advantage.
What does borrowed competence look like in your industry?