By Josh Sadek, Chief Technology Officer, Olympus Technology Services.
Ray Kurzweil has spent decades predicting that machine intelligence will match ours by the end of this decade and surpass it quickly after. For most of that time, the prediction was easy to take lightly. Today, every organisation I speak with is leaning on AI a little more each day, and the train is only getting faster.
That reliance is reshaping how we build businesses. We are redrawing team structures, rethinking ‘roles’, and rewriting our ways of working around teams of agents that draft, code, analyse, schedule, and decide. The skills discussion has shifted. Judgement, communication, persuasion, and the ability to frame a problem well are becoming the premium capabilities, because the machine will happily do the rest.
In the past few months, one question has kept me honest: what happens to our vocational ability when the machine does the rest?
There is an old saying, popularised by the novelist G. Michael Hopf, that hard times create strong people, strong people create easy times, easy times create weak people, and weak people create hard times. It is a blunt tool, but it is worth asking which part of the cycle we are in. My view is that we are entering the easy times. The strong engineers and innovators of the last three decades have built tools so capable that the next generation may never need to struggle the way they did.
We have seen this trend before. Whilst mathematics is still a very real thing, calculators ended mental arithmetic for most people. GPS did not end navigation, but many of us can no longer read a map. A generation of technologists now runs production systems without ever having racked a server. In each case the upside was real and the erosion was slow and quiet.
Now play the tape forward. Imagine an organisation five years from now where a team of agents write the code, run the operations, handle the customers, and reconcile the books. The people who once did those jobs have moved on, shifted their focus to soft or new skills, or retired. The newcomers have only ever supervised and worked with agents. Then something slows the train. Governments are already posing the question of whether frontier AI should be constrained. Energy is a harder limit than policy; the compute behind this revolution is power hungry, and grids in Australia and elsewhere were not designed for it. We have lived through RAMageddon-style shortages in the past, when one supply constraint resets prices across the whole industry. A sharp squeeze on chips, power or cooling would do the same to the AI revolution. And when the agents subside, or become too expensive to run, who in that organisation still knows how to do the work? That is when weak leaders create difficult times.
Perhaps AI designs its own way out, architecting new ways to power and cool itself, and building the bigger, faster factories its growth demands. I think that is likely, but hope is not a strategy. Hoping the machine solves its own constraints is exactly the complacent thinking the old adage warns against.
So my message to fellow leaders is simple. Doing AI is not optional, and neither is understanding it. Deploying an agent is the easy part. Knowing why it works, where it fails, what it costs to run and what happens when it is unavailable is the hard part.
At OlympusTech, we decided the only honest way to learn was to become patient zero. We are running AI through our own business first: our engineers, our service desk, our delivery teams and our back office. We are building the tools, questioning the outputs and keeping the fundamentals alive in our people as we go. So this is an open invitation to come and see what we are doing, challenge our thinking, keep us honest and figure this out alongside us.
