AI as Pedal Assist - Not a Substitute for Control

Revisiting Steve Job’s bicycle for the mind
Everyone has an opinion on AI and the impact it will have on society and our future. I’ve read countless takes on generative AI - some say it’s going to replace us all, others act like it’s a silver bullet.
The common thread? Confidence. Strong opinions. A sense that if you’re not already on board, you’re behind.
I don’t feel certain and I don’t have a sweeping take on where this is all going. When the iPhone was first released in 2007, no one predicted all the ways it is used today.
What I do have is a handful of personal experiences that I keep coming back to and the clearest one is this: Using GPT feels exactly like riding an e-bike.
If you know where you’re going, it gets you there faster
That’s the magic of these tools. If you’ve got a clear sense of direction, AI, like an e-bike, helps you cover more ground in less time. You still need to steer and you still choose the path but the assistance makes the journey smoother.
Where it doesn’t help is when the destination is unclear. AI, like a bike, has no idea where you want to go. It will power you forward, but it won’t tell you if you’re heading the right way.
If you’re not careful, you can end up somewhere quickly... only to realise you’re a long way from where you meant to be.
It’s hard not to use it
Sometimes I’ll be out riding and think, I should just pedal this bit without the motor, push myself a little. And maybe I do for 20 seconds, then I flick the power back on. It’s just so much easier.
It makes sense, why wouldn’t you take the easier path once you’ve seen it?
The same thing happens with GPT. I’ll set out to think through a problem from scratch, and 30 seconds in, I’m reaching for the tool. The ease can be seductive, not because I can’t do it myself, but because doing it alone takes more effort.
There’s nothing wrong with using these tools, but combining them with creativity, judgement, and human touch will always lead to the best results.
More access doesn’t remove risk
An e-bike doesn’t just help you go faster, it also helps you climb higher and access trails, elevations, and terrain you might not reach otherwise. But ultimately, gravity is the most important factor. You hit your highest speeds going downhill, not with power, but with momentum.
That’s where judgement matters most. The assistance gets you up the hill, but it doesn’t help you navigate the way down.
AI works the same way. It opens up new capabilities, but the risk and the responsibility is still yours.
You can’t build a new trail with power alone
An e-bike helps you move faster, but only along an existing path. It doesn’t create something new and it still takes human effort to explore and create something original.
AI is similar. It reflects what it’s been trained on. It can remix and accelerate, but it won’t give you purpose. It doesn’t replace clarity or creativity. If the problem is unclear, no amount of power will get you through it.
Sometimes, we’re still walking
As much as we talk about digital transformation, a lot of work still happens manually. Systems that don’t talk to each other. APIs that don’t exist. Files that still get screenshotted and emailed.
Not everyone’s riding high-end tech. Some of us are still pushing bikes uphill.
And some e-bike users weren’t great cyclists to begin with, hoping a motor would solve everything, only to find they’re still struggling.
The tool doesn’t decide the journey
Steve Jobs described the computer as a “bicycle for the mind.” The metaphor still works, but what he didn’t say is this: The bike doesn’t know where you want to go. That’s still up to us.
AI will evolve to be used in far more ways than it is utilised today. What won’t change is that you’ll still need creativity, clarity, and good judgement to drive it in the right direction.
And for anyone wondering, yes, I did use a GPT to help me write this. The idea was mine, and the tool helped me get here, and here's a real picture of my using pedal assist on my bike, it's definitely easier, but I still have to work.
