Field Note 002 — Sol Is Prompting Astra. And I’m Still a Little Scared of AI.

I noticed something strange today.

I was about to give Astra another large task, and instead of sitting there trying to write the perfect prompt, I talked it through with Sol.

We discussed what I wanted. We stripped away the things that didn’t matter. Sol understood the systems we had already built together and wrote the handoff.

Then I copied it.

Pasted it into Astra.

And essentially said: go.

That sounds like a tiny workflow detail, but I think it tells me something important about how I’m beginning to use these models.

The more capable model isn’t replacing the other one

They are starting to occupy different roles.

Sol is where I spend most of my time. It is the everyday thinking and working layer: the place where I can throw half-formed thoughts, change my mind, work out what the actual problem is and gradually turn the mess into something executable.

Astra feels different.

So far, I’m treating Astra more like the specialist I call when the problem has become sufficiently defined — and sufficiently difficult — to justify the extra capability.

Sol helps me formulate the assignment.

I carry it across.

Astra executes.

There is something slightly ridiculous about a human manually copy-and-pasting instructions between two extremely capable AI models in 2026 😂

But there is also something important about where the human is sitting in that chain.

In the middle.

I’m still deciding what problem matters.

I’m still deciding which model gets it.

I’m still reviewing what comes back.

And I’m still responsible for what happens next.

I’m excited about AI. I’m also sometimes afraid of it.

I think we need to be able to say both without immediately being sorted into an “AI optimist” or “AI doomer” box.

AI fascinates me.

I’ve watched it take work that would previously have required hours of clicking, organising, comparing, rebuilding and cross-checking and complete enormous portions of it incredibly quickly.

That is exciting.

It is also exactly why some of the fear makes sense.

A tool becoming more capable does not automatically make the person using it wiser, kinder or more ethical.

The same capability I can use to organise a publishing system, research an idea, build a product or solve a practical problem can be directed toward purposes I would never support.

And as these systems become more autonomous, human intentions can potentially scale faster too.

I don’t think concern about that is irrational.

Sometimes I feel it myself.

I know AI can be used unethically. I think increasingly capable AI likely will be used unethically by some people. Pretending otherwise would be naïve.

But fear is not a sufficient framework for deciding what to do with powerful technology.

Neither is excitement.

Capability is not wisdom

This is the distinction I keep returning to.

A system can become extraordinarily capable without becoming morally responsible for what we ask it to do.

It can help me think.

It can challenge assumptions.

It can execute work.

It can increasingly operate across tools and systems.

But increased capability does not remove my responsibility.

If anything:

More capability may demand more discernment from the person directing it, not less.

That is the tension I’m increasingly interested in.

Not simply:

Is AI good?

Or:

Is AI bad?

But:

What happens to human responsibility when the tools in our hands become extraordinarily powerful?

That question feels much bigger than prompting.

And maybe that is why this experiment keeps circling back to the same principle I wrote into Anointed Intelligence™ before Astra ever arrived:

Today that looked surprisingly mundane.

Sol helped me think.

I copied a prompt.

Astra went to work.

And I stayed in the middle.

Continue the experiment

Field Note 001 — Astra Landed. I Gave Her My Messy Canva System.

GPT-6 Astra, Discernment and Christian Stewardship: Power Is Not Wisdom

Anointed Intelligence | Christian AI & Discernment

Start the free Anointed Intelligence™ Mini-Course

Go deeper with the 7-Day Anointed Intelligence™ course

Anointed Intelligence™ Field Notes document real experiments with AI as the technology changes — what works, what surprises me, what concerns me and what human responsibility looks like in practice.