BIBA says talent beats AI. Ask which one lets the other do its job.
A trade body's own survey just proved the industry's biggest AI problem sits at board level, not on the front line.
Talent attraction and retention topped BIBA's 2026 member survey at 40%, ahead of AI and automation at 24%. On the face of it, a sensible answer from an industry short of people: 72% of UK brokers are actively hiring, graduate vacancies fell 18% in 2025, and only 4% of young people say they fancy a career in insurance. Faced with numbers like that, of course talent feels like the more urgent problem.
Look again at what the survey actually asked, though, and a different story emerges. Members were asked to rank AI against talent as if the two were competing budget lines — two boxes on the same list, one likely to be ticked at the expense of the other. That a trade body's own membership answered the question as posed, without anyone flagging that the premise was wrong, tells you less about broker priorities than about how AI is understood at the top of this industry.
A principal who genuinely understood how AI gets implemented in a brokerage — which is an operating-model question, not a procurement one — would never rank it against talent, because they'd know the two aren't substitutes. AI doesn't compete with people for a place in the business. It competes with time: specifically the hours a stretched handler currently loses to renewal admin, chasing documents, drafting standard emails and re-keying the same data into three different systems. Free that time and you haven't replaced the handler. You've made the handler worth more, and made the same headcount capable of carrying a bigger book.
That distinction sounds obvious once stated. That most respondents to a well-run industry survey didn't state it themselves is the finding actually worth sitting with. It isn't a resourcing debate. It's a governance gap wearing a resourcing debate's clothes.
The cost of that gap shows up in both directions. Boards that see AI as a rival to headcount under-invest, and keep paying more each year for the same bottleneck. Boards that don't understand it at all over-invest in the wrong shape of tool — buying something to tidy up work a human already did, rather than building the sequence the blueprint actually requires: AI performs the work, a human verifies it. That's not a subtlety. Get that sequence backwards and you've bought expensive software that makes your existing process feel busier without making it faster.
There's a sharper version of the same problem sitting just behind it. A board that can't distinguish a simple, single-response AI workflow from something genuinely agentic — a system that runs on its own trigger and decides what happens next without a person starting each instance — can't sensibly answer the question that actually matters: are we ready to let this run with less oversight, or does it still need a human release gate. That's precisely the judgement Consumer Duty and SM&CR expect a senior manager to exercise personally. It isn't a decision you can delegate to whoever signed the software contract.
This is why the talent-versus-AI framing is backwards at the level that matters most. The scarce resource in UK broking right now isn't headcount, and it isn't AI licences. It's board-level fluency in how the two actually interact — fluency that lets a principal look at a renewal-admin bottleneck and correctly diagnose it as an automation candidate rather than a hiring gap, and look at a claims-triage proposal and correctly diagnose it as something needing a human sign-off gate rather than a black box to switch on and forget.
Firms that build that fluency at the top table won't experience talent and AI as a trade-off, because they won't be choosing between them. They'll be sequencing them — using automation to make the people they've got, and the people they're competing hard for in a shrinking graduate pool, worth considerably more than a competitor's equivalent hire. Firms that don't will keep answering surveys like BIBA's the way most of the membership just did: treating a design question as a spending question, and arriving at the wrong answer for entirely reasonable-sounding reasons.
The next survey worth running isn't "talent or AI." It's "does your board actually understand what AI implementation means for this business" — and going by this one, most broking boards currently couldn't answer it.