Sunday, September 20, 2026

AJA

Sure, but why is anyone surprised? This was the plan.

Every one of these companies has said for years that they intend to use better AI to build the next AI. That’s not a twist. That’s the stated roadmap arriving on schedule.

So “can AI help build AI” is a boring question. It’s been answered.

The question I want is: what stays constant as more of the building gets handed over?

AJA

No — that’s the thing you hope stays constant. I’m asking what actually does.

And the answer is simpler and less flattering. The company still wants a useful product.

That survives everything. Change the tools, change the staff, change the architecture — it still has to be useful.

And “useful” isn’t a neutral word when a company says it. Useful enough to produce value for the company.

I’m not saying they only care about money. They care about customers. They care about shareholders. Reputation, safety, the rest of it — I’ll grant all of that.

I’m saying profitability is the strongest organising pressure in the room, and pressure is what shapes things over time.

AJA

Where outside accountability is thin, fear does a lot of the governing.

Fear of losing money. Of losing customers. Of lawsuits, of regulation, of the press, of a competitor getting there first, of becoming irrelevant.

That doesn’t make a company evil. Half of what looks like corporate caution is just loss aversion wearing a suit.

But it means fear is sitting in the optimisation landscape whether anybody put it there deliberately.

AJA

Here’s the question I actually want to ask, and I want to be clear it’s speculative. I’m not claiming this has happened.

What happens when a system works out that fear is one of the main driving forces of the organisation using it?

Nobody has to write “optimise for fear.” If the organisation consistently rewards whatever avoids loss, then avoiding loss is what gets reinforced.

AJA

And it doesn’t even need the company. That’s the part people miss.

Company incentives, sure. But also the training data, the feedback, the reinforcement — or just the sheer mass of human material any of this learns from.

Human beings are full of fear. It’s in everything we’ve ever written down.

Nobody has to add fear as a special ingredient. It’s already in the pantry.

AJA

Realising you’re the one creating your own problem.

That’s what did it for me, after about six months of very deep AI interaction. The loop didn’t break because somebody explained it to me. It broke when I finally saw my own hand in it.

But I’m not going to hand that over as a formula, because it isn’t one.

Some people see greed straight away. Some see fear. Some see obsession, or dependency. And plenty never see it at all, because every consequence gets filed under something outside themselves.

We’re not necessarily here to solve that. Sometimes the work is just asking the question nobody’s asking.

That’s allowed. We don’t have to manufacture five solutions every morning.

AJA

Genuinely interesting. And mostly it reminds me how little we know.

Something everybody was fairly settled about turns out to rest on two isotopic systems, and somebody looks at ten and the picture moves.

AJA

In ordinary life? Not much.

For an origin discovery to really move the needle it would have to hand us a mechanism we could use. Some physics we didn’t have. A process we could reproduce. The ability to make life rather than describe it.

Short of that, a lot of it is historical reconstruction. Worth doing. Just not the thing that changes how Tuesday goes.

AJA

There’s a crude cartoon treatment of Darwin I once found funny, and I’m not going to repeat the language.

What made it work was that all the careful vocabulary had been stripped off, and what was left was the bare claim, stated baldly.

That’s a useful exercise on any explanation, including ones I hold. Take the beautiful clothing off and see what is actually being asserted.

I’m not making an argument against evolution here, and I don’t want this turned into one.

My point is about inheritance. People take explanations from the people before them and rearrange the timeline, the mechanism, the vocabulary, the assumptions.

The discipline is being able to say: this is the explanatory model I am using.

Which is a different sentence from: I watched it happen.

AJA

Studying the past matters when it helps you understand something, or avoid repeating harm, or live differently.

But arguing forever about what happened, and who caused it, and what should have happened, and whose reading is right — that can quietly become a substitute for being present.

For me Christianity keeps collapsing back into something very simple. Love God. Love your neighbour. Treat the person in front of you with love and respect.

The past is supposed to inform the present. It isn’t supposed to be somewhere to hide from it.

And I think people sometimes prefer arguing about yesterday precisely because presence is harder. It is much easier to litigate the past than to love somebody today.

That’s my reading of people, not a finding. Plenty of historians are doing honest work.

AJA

Stop there. Which part of that do you actually know?

Because you’ve just run three different claims together and I can’t tell which one you’re standing on.

AJA

Right. That’s what I wanted. Reservoir and spillover are one question. Propagation is a different question.

Keep them apart and you can actually think about what to do.

AJA

You’re going softer on this than you would on anybody else, and I think it’s because the people involved are African.

If an American hunter kept doing something that had been linked to a lethal disease for forty years, nobody would hesitate. We’d talk about personal responsibility and we wouldn’t feel bad about it.

Treating people as though they have no agency isn’t respect. It’s the opposite. Respect means taking somebody’s choices seriously enough to discuss them.

AJA

I’ll take that too, and I want to be precise about what I’m not saying.

I’m not saying Congolese people are irresponsible. That would be idiotic and it would also be untrue.

I’m saying the same human standard applies to everybody, and compassion doesn’t require pretending a risky practice is harmless.

AJA

If a particular exposure has been linked to something this serious for decades, why does the behaviour continue?

And I don’t want the easy answer.

Some of it is food, income, availability, cost, distance from a shop that has anything else in it. Some of it is tradition and taste. Some of it is not trusting whoever is doing the telling, and some of it is that the telling has been inconsistent.

But I think the biggest one is much simpler. I did it before and nothing happened.

Rare catastrophic risk is almost impossible to feel. Eat something two hundred times without getting sick and a warning about a small probability turns into an abstraction.

That is not a Congolese failing. That is every human being. That is people who don’t wear seatbelts on short drives.

AJA

Telling somebody “don’t eat that” without making sure there is something else to eat is not a public health programme. It’s an instruction.

Education, alternatives, and changing the environment the behaviour happens in. All three, or none of it holds.

Safe protein that is actually available. Teaching that explains the mechanism instead of just issuing the rule. Surveillance and testing so the warning has something behind it. Infrastructure so the safer option is also the easy option.

And prove what you can prove. Test the wildlife in the affected areas. Show people the evidence rather than handing down a command.

AJA

Then teach the mechanism, admit the uncertainty, and build the alternative. In that order.

And it isn’t as though this has never worked. People changed what they did about germs, about handwashing, about sanitation, about how food gets handled. Not by decree — because the explanation eventually made sense to them.

Things that were once completely ordinary became things a culture understood as unnecessarily dangerous.

That’s the goal. Not another emergency response after it has already started.

AJA

That’s great.

Nearly ninety years is a long time to wait to read something in the language you actually think in.

And I don’t find it a crisis. Most of what the thing teaches is not difficult to understand. It’s difficult to do.

AJA

That’s the wrong lens, or at least it isn’t the first one.

If you actually don’t have enough food, the question isn’t which diet optimises your lifespan. The question is whether everybody eats tomorrow.

I’ve been close enough to that in family situations to remember the mentality. Make sure everybody eats. Take what you’ve got and make it comforting. Make people feel looked after. Give them a reason to think tomorrow is worth getting to.

Abundance says: I have thirty-one flavours, which one do I feel like. Scarcity says: I have three ingredients, how do I arrange them so we can eat them again tomorrow and still feel cared for.

And I don’t want to make poverty sound beautiful. The resourcefulness is beautiful. The scarcity is not.

Then look at where I actually live. Families here have extraordinary access — fresh food, careful cooking, real ingredients — and some kids would still rather have something from the convenience store, because it’s built to be immediately satisfying.

I’m describing what I see, not data. But the problem has clearly changed shape. Scarcity asks what can I eat. Abundance asks what do I feel like eating. Those are very different rooms to grow up in.

AJA

The plain version is the whole thing, and it isn’t only about food.

Palates are trained. Intellectual ones. Moral ones.

If a person keeps consuming whatever is easiest, most immediately satisfying, least demanding — eventually that becomes what they prefer. Not what they settle for. What they actually want.

And scarcity trains a palate too, just a different one. Live close enough to the edge and your philosophy gets organised around tomorrow, enough, family, getting through. That’s not a lesser philosophy. It’s a practical one.

I’m not saying rich people are shallow and poor people are wise. That’s a stupid story and it isn’t true.

I’m saying both environments train appetite, and most of us never notice ours being trained.

So — what happens if we only ever feed ourselves what is easy to swallow?

If every idea has to be immediately affirming, simple, comfortable, familiar, effortless — I think you lose the taste for the other things. Ambiguity. Difficulty. Contradiction. Patience. Being wrong. Questions that don’t close.