Recently I have spent quite a bit of time thinking about AI’s role in organisations, from leadership and board decisions to hiring strategies and global impact. (I didn’t say I figured it all out, just that I have been thinking about it.) But in addition to reading and writing about AI, I’ve used it a ton. Far more than I expected (daily use was not on my bingo card a year ago, but here we are).
Like everyone, I was initially mesmerised by the super robust, well-written, thoughtful-sounding responses. It was like the stuff of 1950s science fiction had finally arrived.
Eighteen months ago: I barely glanced at ChatGPT. Today, I’ll toss almost anything into AI ‘just to see’ what it comes up with. Fine. This is the new reality. But what’s even more interesting (to me) is now it has become almost pedestrian. Not quite boring, but I use it less like an oracle and more like a tool. A good one. A fast one. But it has become an input to my decision-making rather than leading it.
All of this has meant I can also quality test what I get back – I come with my own expertise and experience in hand, and if AI gives me things I feel strongly aren’t the right solution, I redirect.
The steady state of AI might be when we learn to leverage the tool and moderate the reliance.
Organisations are rightly still working through how AI will be fully leveraged. And AI is still at such an early stage that that answer will continue to evolve. But I think the real next step is: how do people continue to drive strategy and judgment once AI has settled into the infrastructure? It feels like a lot of companies are still trying to lean into AI and test its limits.
But I do think most leadership teams still have room to pivot to folding AI into place as a tool — like a spreadsheet replacing a calculator, not replacing the CFO. And that’s not nothing — a spreadsheet can do a lot more than a calculator.
But no one ever expected a spreadsheet to decide which market to enter, or which hire was next. It made the modelling faster, the scenarios cheaper to run, the errors easier to spot. Then a person, sitting with incomplete information and real consequences had to rely on mere human synthesis. . . and make a decision. Tools have always been built to scale tasks and we have not yet built one to truly replace human judgment. Not yet.
AI produces well-written ideas on demand, in unlimited quantities, for everyone, at once. Which is precisely why it can’t be the differentiator. The AI answer will become standardised. What’s scarce is the thing underneath — the human judgment that knows which polished answer is actually right for that business, at that time.
We’ve seen a version of this before. In the 1920s, researchers at Western Electric’s Hawthorne Works ran experiments on factory lighting. Brighten the floor, and productivity rose. Dim it, and productivity rose again. But also, dim it nearly to darkness, and productivity still rose.
In other words, it wasn’t the lighting. (Years later it was learned that the variables were social pressure and paycheck timing. Sooo…nothing has changed.)
AI is transforming how humans do things. But it’s not (yet) transforming humans themselves.
I suspect we’re running that experiment again with AI as the lighting. Switch it on across an organisation and everyone’s output improves: faster drafts, sharper-looking analysis, that universal polish. It’s tempting to call that transformation due to AI.
But when everyone’s numbers move together, the lift tells you very little about who is actually performing. The durable (and human) drivers sit beneath who displays better judgment, who builds trust, whether people feel safe enough to raise a problem while it’s still small. Which is the stuff those of us in talent management and organisational psychology have been analysing all along.
The companies that get to the steady state fastest — AI folded into the infrastructure, unremarkable, load-bearing — will be the first to see clearly again. And what they’ll see is that the strategic questions have quietly returned to where they always lived: with people. That is still who develops judgment in your leadership pipeline. You still need leaders to make the judgment call, even if based (in part) on the ever more sophisticated tools that inform them.
AI will keep transforming how work gets done. That’s more or less settled. Personally, I think AI will ultimately change our lives…maybe at the scale of electricity. For now, however, AI doesn’t have to transform human performance. Because performance was never really about the conditions. The lights went dark; the numbers kept climbing. The people remain the strategy.
