How AI is changing the balance of power between people and institutions

The moment I realised I was turning into Batman, but with a pen, was when I realised I was effectively being gaslit by InPost.

They’d lost a parcel I’d sent, and I spent weeks arguing with them about the insurance I’d paid good money for. They wouldn’t pay out because there was tobacco in the parcel. It got nuts at one point, and they told me they’d destroyed it. I remember asking something like, well, what do you mean? Are you saying you burnt it? Did you just smoke it?

To give you an idea of how crazy it got, there was one particular day where I received an email from their customer service team saying they couldn’t find the parcel and I should claim on the insurance. This was after weeks of going around in circles. Around the same time their legal team were refusing the claim on the grounds that tobacco was a banned item. Then there was a knock at the door and, sure enough, there was the parcel.

I think all of that happened within the space of a couple of hours.

It had basically been like that for weeks. Different answers from different parts of the same organisation, positions changing, things I’d already explained coming back round again. After a while the problem stops being whether you’re right. The problem is that you have to keep the entire fucking thing in your head.

Every time somebody replies you have to reconstruct the case. What happened when? Who said what? Where was the email where they said the exact opposite? What did the terms actually say? Is the thing they’re now claiming even relevant? Then there are the calls, the chasing, the research and the next letter, followed by remembering to chase that when nobody replies.

For the organisation this is normal business. For you it is eating your Tuesday afternoon.

That is quite a significant imbalance of power, and I don’t think I’d properly appreciated how much of it was administrative until this happened. Organisations have systems, records, specialist departments and people whose job it is to manage this stuff. The person on the other end usually has an inbox, a memory and a finite amount of patience.

Then I realised I didn’t have to manage it that way any more.

I could use AI.

I started developing a methodology around disputes like this. Telephone calls can be recorded and transcribed. Emails can be fed into the same case history. The AI can keep track of the chronology, compare what different people have said, surface contradictions, help research the relevant law, draft letters and keep track of what needs to happen next.

The useful thing is not really any individual part of that. It’s that I don’t have to keep reconstructing the dispute myself.

If I vaguely remember somebody saying something on the phone three weeks ago, I can find it. If I think the company’s position has changed, I can compare what they’re saying now with what they were saying before. If I need to understand whether an argument they’re making has any legal basis, I can get the system to research it and point me towards what I need to check.

I’ve now got this to the point where I can take a dispute all the way through to a court application with surprisingly little administration on my part. I think I could probably use the same system to help prepare litigation as well, although I haven’t tested that yet. Certainly most of the machinery is already there.

And somewhere in the middle of building all this I realised I was becoming Batman, but with a pen.

Batman is interesting because he doesn’t actually have any superpowers. He is basically a bloke who has made himself disproportionately capable by collecting tools, information and skills, then working out how to use them together.

Obviously I have fewer grappling hooks.

But the underlying idea is surprisingly similar. AI lets an individual assemble capabilities around themselves that previously belonged mostly to organisations or people with money. Research, transcription, record keeping, drafting, analysis, administration. None of those things is especially impressive on its own, but put them together and suddenly one person can operate at a completely different level.

And that changes more than convenience.

A lot of bad institutional behaviour works because resisting it is expensive. Not necessarily financially expensive, although sometimes that too. It costs time, concentration and emotional energy. It requires you to stay organised for weeks or months while getting on with the rest of your life.

The organisation doesn’t necessarily have to prove you wrong. It only has to make proving yourself right irritating enough that you eventually decide it isn’t worth it.

Send another form. Refer you somewhere else. Ignore half the question. Ask for information you’ve already supplied. Wait long enough between replies that you have to remember what the argument was about in the first place.

Eventually most people have something better to do.

The interesting thing about AI is that it can remove quite a lot of that cost.

As a relatively sophisticated person this stuff is fairly easy for me. I understand systems, I’m comfortable with technology and I’m bloody-minded enough to keep pulling at a contradiction when something doesn’t make sense.

But once I’d worked out what was possible, that started to bother me.

There wasn’t much point in simply becoming extremely good at winning my own disputes. I could make myself harder to push around, which is useful, but it doesn’t change the underlying problem.

So I’ve spent the last few weeks figuring out what to do with it.

I’ve ended up with two things.

The first is that I’m going to publicly call out bad behaviour when I see it. Not every mistake or badly handled customer service interaction. People make mistakes and organisations are messy. I mean the stuff where the system itself starts depending on confusion, contradiction or exhaustion.

The second part matters more. I’m going to publish the methods I use to deal with it.

The process, the prompts, the way I capture evidence, how I maintain the chronology, how I use AI to interrogate what I’ve been told and how I move from a vague sense that something is wrong to a case that can actually be pursued.

Because there is a much bigger opportunity here than me becoming unusually difficult to fuck with.

For a long time the organisation had the machinery and the individual had persistence.

Now the individual can have machinery too.

If enough people learn how to use it, some forms of bad behaviour simply stop working.

The useful thing about Batman’s utility belt isn’t that Batman has one.

It’s that now we all can.