Who's flying the plane?

AI takes some skills off you and hands you others. The catch is choosing which ones to keep.

On the night of 1st June 2009, an Airbus A330 took off from Rio de Janeiro bound for Paris, climbed to its cruising altitude over the Atlantic, and a few hours later fell into the sea. All 228 people on board were killed.

The aircraft was still flyable. Its engines were running, its controls were responding, and its structure was intact. What happened was that a cluster of small sensors on the fuselage iced over in a storm.

The airspeed readings scrambled for less than a minute, and the autopilot, which relies on those readings, did the sensible thing and switched itself off, handing the aircraft back to the two pilots at the controls.

They had flown countless times. What they had rarely done was fly a large aircraft by hand, at altitude, in the dark, with the automation gone and the instruments contradicting each other. One of them pulled the nose up and held it there. The A330 climbed, bled off speed, and slipped into a stall it never came out of.

It took three and a half minutes to reach the water, and for most of that time the crew did not grasp that the wing had simply stopped flying.

When investigators reconstructed it, the aircraft had remained capable of flying throughout. Plenty had gone wrong to get them there - confusing alarms, a stall warning that came and went, a crew struggling to read what the plane was telling them.

But they were also pilots whose opportunities to practise manual flying at altitude had become vanishingly rare - and with them the instinct you need when the machine hands the controls back.

The trade

There’s a word for a skill wasting away through lack of use, and it’s one I keep hearing recently. Atrophy. It’s what happens to a muscle you stop using - it doesn’t disappear, it just weakens, until the day you try to use it and it just doesn’t respond.

The people I hear using the word aren’t pilots this time. They lead teams, run businesses, do knowledge work for a living. What they’ve noticed is that the more they hand to AI, the less certain they are they could still do the thing themselves. It gets done, and faster, but some part of the ability has diminished.

And I understand the worry. I recognise certain skills and behaviours I used to lean on that are undoubtedly weaker now.

But when I look honestly at my own work, atrophy only describes half of what’s happened. Because there are also things I can do now that I simply couldn’t a couple of years ago. Not least, get properly underneath a subject I knew nothing about a fortnight ago - aviation safety being the obvious one.

So what’s going on is a trade - some abilities weaken while others - ones that were out of reach before - come into range.

Which means there’s a better question here than whether or not AI weakens your skills. Because, yes, it definitely weakens some of them. The more important point is whether you get any say over which.

Some of what fades, you won’t miss. The problem is you’re not the one choosing what goes.

The slow way

Here’s the part I’m genuinely unsure about.

The old way I learned things was slow. I’d read the thing, wrestle with it, get it wrong, go back and do it again by hand. It was inefficient, it took ages - but the lessons stuck. What I learned that way, I still have up there.

Now I learn differently.

I go round and round on a problem with a machine - testing an idea, throwing it out, reshaping it, again and again. It’s quicker, but it often feels like I’m wasting my time - circling the same ground instead of getting anywhere.

The old way contained a lot of friction. Some of it was pointless. But some of it was the mechanism by which the lesson became mine. AI removes both kinds at once.

Slow or fast isn’t the thing that matters. What matters is whether I’m doing enough of the thinking for the idea to lodge in my own head, or just nodding along when the machine hands me a good answer.

I can’t tell yet which I’m doing.

Aviation’s response is one part of this where somebody made a deliberate choice. They didn't tear the autopilots out - the automation was saving far more lives than it ever cost, and everyone knew it. They changed the humans instead.

Pilots started being deliberately exposed to the kinds of conditions that had overwhelmed that crew - unreliable airspeed, high-altitude stalls, the automation vanishing without warning - rehearsed until the right response was there when it mattered.

They picked the skill they refused to let waste, and built the practice to keep it alive.

Hardly anyone handing work to AI is making that choice on purpose. The skills that weaken are simply whichever ones we happen to stop using first - not the ones we’d fight to keep if we stopped to think about it.

So the real question has nothing to do with speed. Of everything you’ve started handing over, which is the one skill you’d never want to lose?

Thanks for reading.

Ollie

Ollie on Work is a weekly newsletter about what I’m learning from building with AI, advising leadership teams, and trying to bridge the gap between what technology can do and how businesses actually work. If someone forwarded this to you, you can subscribe here: