A colleague was recently told exactly that. It is a bold statement designed to provoke a reaction, but it misses a more important point.
When I started on a trading floor in the 1970s, we were already working with technology that, in today’s language, might be described as an early form of AI.
I worked for a firm that traded precious metals and had links to London’s daily gold fix. A brilliant engineer, Thomas Peterffy, who would later found Interactive Brokers, developed a system that identified arbitrage opportunities across markets. It could detect when gold could be bought in New York and sold in London profitably or highlight pricing discrepancies across a range of commodities. When an opportunity appeared, it flashed onto the trader’s screen.
There was no electronic trading. A trader would have a phone to each ear, executing both sides of the transaction as quickly as possible. Upstairs were rows of machines requiring constant cooling, an entire room of computing power dedicated to finding opportunities. Looking back, it feels like a distant ancestor of today’s AI data centres.
We didn’t call it AI, we called it work.
The machine found the opportunity and the human decided what to do with it.
That is why the idea that someone should retire if they don’t “embrace AI” misses the vital point. AI, in one form or another, has been part of financial markets for decades. Today’s large language models are remarkable in their scale and accessibility, but the underlying principle has not changed. Technology identifies possibilities but people apply judgement.
When an AI tool produces a first draft, summarises a document or analyses a dataset, that is valuable. The real work begins afterwards. Professionals still need to ask what it may have overlooked, what assumptions it made and where it might fail in the real world.
In practice, that means:
- Let the system surface the opportunity and the professional make the decisions.
- Keep hold of the question. Tools generate answers, but professionals decide which matter.
- Understand what sits beneath the output: the data, the assumptions, the controls and the governance.
- Expect variables and inconsistency. A model that performs brilliantly one morning may struggle with a slightly different problem that afternoon.
- Prefer the dependable system that quietly delivers over the impressive demonstration that fails under pressure.
None of that is especially glamorous. Back on the trading floor, when an arbitrage opportunity appeared and we captured it, nobody celebrated. We simply moved to the next trade. The value was never in the machine performing a clever trick. It was in the partnership between technology and experience.
That still remains true today.
Clients don’t pay because someone knows the name of the latest model. They pay because reliable outcomes are delivered: licensing that clears, structures that settle and controls that stand up when tested.
Experience matters because it provides perspective. It stops the same mistakes being made twice. The people who sound most cautious about new technology are often the people who built the foundations the industry now takes for granted. They have seen elegant models fail in messy markets. They understand how incentives distort behaviour. They know that something which works 99% of the time can still cause serious problems on the wrong day.
That’s not resistance to technology; it is professional discipline.
Curiosity matters just as much. If AI can remove the first 60% of repetitive work, let it. Use the time gained to focus on the part that still depends on human judgement: framing the problem, weighing the risks, recognising the exceptions and knowing when “almost right” is still wrong.
So no, failing to be enthusiastic about AI isn’t a reason to retire.
The real test is the same as it has always been which is that we need to learn the tools, understand their limits, apply judgement and use technology to produce better outcomes.
The machines may have become smaller and infinitely more powerful, but the relationship hasn’t changed.
The machine assists, but the human remains responsible.
– Rosemarie Connell, Senior Managing Director, Integrated Solutions