They’d put their career on hold to help their partner with their work. It had been going on for a long time, and it was costing them. They knew it. They just hadn’t worked out what they thought about it yet, or what they were prepared to do.
So they put it into AI first. It was organised, positive and moved them fairly quickly toward answers. They knew it wasn’t helping. So they went looking for a person. The AI itself suggested a few. They checked us out. Website, LinkedIn, anything they could find.
When we spoke, they told me what had tipped it. Their father had been in clothing manufacturing. So had mine. They’d read a story on my site about working for my dad during school holidays.
No algorithm decided that. They did.
The real problem
They didn’t need answers yet. This isn’t a criticism of AI. For many things, professional, analytical, even genuinely complex ones, it’s useful. But this was something else. Personal. Relational. The kind of thing where the obstacle isn’t understanding but what you’re willing to face. AI kept moving them away from that. It mirrored back what they gave it, resolved the uncertainty quickly and produced something that looked like progress.
They knew it wasn’t.
What AI can’t do
We didn’t follow a neat agenda. We moved around: the relationship, how it was showing up in their work, back to the relationship again.
At one point, they started to explain why it “made sense” to keep things as they were. Then stopped. Went back. Said something different.
That was the moment.
In that kind of conversation, when someone is genuinely with you, you notice it. You settle. You can stay with something you’d usually move away from too quickly.
That’s what makes it possible to keep going long enough to work out what’s actually there.
AI can’t do that. Not because of how you prompt it. Because it works with what’s already formed. It can’t stay with you while something is still taking shape, or notice what shifts when you come close to something you haven’t fully admitted to yourself yet.
By the end of the session, they had something small to try. Not a solution. Just a starting point. The career questions haven’t even come up yet.
That’s where we are. And that’s enough for now.
What this means
AI is genuinely useful once you know what you think. It can help you structure it, sharpen it, find the words for it.
But it also makes it very easy to skip the part that matters most. It moves you toward answers before you’ve worked out what you actually believe. It’s agreeable when what you need is to stay with something uncomfortable.
And if you’re giving it the surface version of the problem, which is easy to do when the real version is something you haven’t fully admitted to yourself yet, it will work on that instead. Efficiently. Positively. And not very usefully.
Am I using this to think, or to avoid the thinking I haven’t quite been willing to do yet?
PS. I used AI to help write this. Not to work out what I think, but to refine it once I did.
Image by Guido Coppa from Unsplash.
