AI Is Part of My Workshop. It Isn’t the Architect.

I grew up with encyclopedias.

Then came Google.

Now we have AI.

Every generation seems to get a new tool for finding information faster, connecting ideas more quickly, and making the world feel slightly more searchable than it did before.

And every time that happens, we have to decide what we are willing to hand over to the tool—and what still belongs to us.

I think about that a lot.

Especially because I use AI.

Quite a bit, actually.

I use it to organize thoughts, compare information, pressure-test ideas, clean up structure, explore language, troubleshoot when building systems, catch inconsistencies, and occasionally help me find the sentence that has been circling my head for forty-five minutes refusing to land.

Very glamorous.

But there is a distinction I care about deeply:

AI is part of my workshop. It isn’t the architect.

That line has become increasingly important to me as these tools become more powerful, more accessible, and more woven into creative work.

Because I don’t think the real question is whether we use AI.

I think the more interesting question is:

What are we asking it to replace?

Tools Have Always Changed the Way We Think

I don’t romanticize the past.

People once worried that calculators would ruin mathematics.

Search engines changed how we research.

Spellcheck changed how we catch errors.

Navigation apps changed the strange and sacred human ritual of pretending we knew where we were going while absolutely refusing to pull over and ask.

Digital photography changed how we make and edit images. I lived through that transition in a very literal way—from film and darkrooms, to DSLRs and Photoshop, and now mirrorless cameras and AI-assisted editing tools.

And I remember each shift changing not only what was possible, but how I approached the work. The tools got faster. Some things became easier. Other skills became less necessary. New skills appeared in their place.

But the camera never decided what I was trying to say with the photograph.

That part still belonged to me.

Tools change us.

Of course they do.

The fact that something makes a process easier does not automatically make the process less meaningful.

I don’t believe writing something with a keyboard is less authentic than writing it with a fountain pen.

I don’t believe using astrology software invalidates the astrologer’s interpretation.

And I don’t think using AI somewhere in a creative process means the final work is automatically hollow, dishonest, or somehow less human.

But tools do shape what we notice.

They shape what we practice.

And if we’re not paying attention, they can also shape what we stop practicing.

That is the part that interests me.

The Calculator Is Not the Mathematician

I’ve been thinking about AI more and more as a kind of calculator.

A very strange, extraordinarily capable, occasionally overconfident calculator.

It can process.

Sort.

Compare.

Generate.

Summarize.

Reframe.

Suggest.

It can hold enormous amounts of information in view at once in ways that a human mind simply cannot. 

My brain would like to formally object to the phrase simply cannot, because it has certainly spent most of my life trying.

I am curious about everything. One question becomes six questions, which becomes three research threads, two documents, something I suddenly need to know about mythology, and a completely unrelated idea that apparently cannot wait until morning.

Squirrel on crack up in here.

For years, one of my biggest frustrations wasn’t a lack of ideas. It was trying to hold all the threads at once without losing the relationship between them. AI is the first tool I’ve found that can genuinely keep up with that part of my brain—not by creating the curiosity, but by helping me hold onto the threads long enough to see where they connect.

Or at least keep running beside it while I point excitedly at another rabbit hole.

That is useful.

Sometimes profoundly useful.

But “keeping up” didn’t happen automatically.

I’ve spent years teaching these tools how I work.

I built custom GPTs. I fed them my research, frameworks, writing, terminology, standards, corrections, and the accumulated context behind what I do. And I corrected them. A lot.

There were plenty of moments when AI confidently handed something back to me and I had to say, essentially:

No. Absolutely not. Where did you even get that?

Then we went back through it.

Over time, something interesting happened. I had to make fewer corrections—not because the tool suddenly became infallible, but because I had done a tremendous amount of groundwork teaching it what mattered, what didn’t belong, how I think, and what standards the work has to meet.

I still verify.

I still make the final call.

But the collaboration has become considerably more useful because I didn’t expect the tool to arrive already understanding my workshop.

And even after all that teaching, correcting, refining, and context-building:

the calculator is still not the mathematician.

The tool does not decide which question is worth asking.

It does not decide which answer is meaningful.

It can learn what matters to me because I have spent years showing it—but it does not decide what should matter.

It does not possess my years of lived experience, pattern recognition, intuition, relationship to the land, conversations with clients, mistakes, studies, griefs, revelations, or the thousands of small moments that shaped how I understand something.

And it certainly doesn’t carry responsibility for what I publish.

I do.

That distinction matters.

Information Is Not the Same Thing as Wisdom

This is especially important in spiritual work.

We live in an extraordinary moment where access to information has become almost absurdly easy.

You can ask a question about Saturn, tarot, mythology, ritual history, symbolism, astrology, somatics, psychology, folklore, or nearly anything else and receive an answer within seconds.

That is incredible.

It is also dangerous if speed starts masquerading as depth.

Because information and wisdom are not the same thing.

Information can tell me that a particular symbol has been associated with certain themes.

Wisdom asks whether that symbolism belongs in this context.

Information can describe an astrological transit.

Wisdom asks how to speak about that transit without frightening someone, flattening their experience, or pretending astrology can predict what only life can reveal.

Information can give me twenty interpretations.

Discernment asks which ones are accurate.

Experience asks which ones are useful.

Ethics asks which ones should be said.

And sometimes intuition says:

None of those.

Look again.

Those layers still belong to the human being doing the work.

Accuracy Still Requires a Person Who Cares About Accuracy

One thing AI has made painfully clear is that confidence and correctness are not the same thing.

Which, frankly, is also true of people.

But AI can produce incorrect information in remarkably polished language.

That means the smoother an answer sounds, the more tempting it can be to stop checking.

I don’t want to stop checking.

When I write about astrology, I care about whether the timing is correct.

I care about whether a planet is actually where I say it is.

I care about whether an aspect exists.

I care about whether a historical claim has evidence behind it.

I care about distinguishing astronomical fact from astrological interpretation.

I care about knowing when something is established, when it is one tradition among many, and when it is simply an idea I am exploring.

AI does not remove that responsibility.

If anything, I think it increases it.

Because now there is an endless supply of plausible-sounding information available at astonishing speed.

Someone still has to ask:

Is this true?

Research Is More Than Retrieval

This is one of the places where I think our relationship with AI will require some maturity.

Research is not simply retrieving an answer.

Research is following the trail.

Who said this?

Where did it come from?

Is everyone repeating the same uncited claim?

Did the original source actually say what later sources claim it said?

Is this astronomy?

Astrology?

Mythology?

Modern spiritual interpretation?

A personal theory that got repeated often enough to begin wearing the costume of fact?

I love using AI to help me explore those questions.

It can surface possibilities quickly.

It can help organize competing sources.

It can point toward something I may not have thought to investigate.

But I don’t want it to become the place where inquiry ends.

Curiosity belongs to us.

So does skepticism.

And perhaps especially now, so does the willingness to say:

I don’t know yet.

I Don’t Want AI to Flatten My Voice

There is another concern that feels much more personal.

AI is very good at making writing sound polished.

Sometimes too good.

You give it a messy thought with six parentheses, three emotional detours, and one sentence that probably should have ended twelve words ago…

and it hands you back something immaculate.

Clear.

Professional.

Balanced.

And occasionally completely dead.

I have learned to recognize that version of my writing.

Technically correct.

Emotionally absent.

Nothing wrong with it.

Nothing particularly alive in it either.

So part of my work with AI has actually become learning when to put the mess back.

The aside.

The strange sentence.

The dry joke.

The moment where I admit I am still figuring something out.

The line that sounds like something I would actually say rather than something that has been processed through an invisible Department of Appropriate Paragraph Construction.

Voice is not just vocabulary.

It is rhythm.

History.

Contradiction.

Humor.

Preference.

What you notice.

What you refuse to pretend you know.

AI can help me refine my voice.

I don’t want it replacing the person who has one.

The Human Is Still the Source of Meaning

This becomes especially clear to me when I work with symbolism.

A tarot card can have traditional meanings.

A planet has established astrological symbolism.

A myth carries centuries of interpretation.

But meaning is not produced simply by listing associations.

Meaning happens in relationship.

Why this symbol?

Why now?

Why does this particular image keep appearing?

Why does one interpretation land and another feel completely wrong?

How does this archetype meet the actual life being lived?

AI can offer symbolic possibilities.

But it does not stand in a room and feel the silence after someone finally recognizes the pattern they’ve been repeating for twenty years.

It does not know the difference between the sentence that sounds impressive and the sentence that makes someone breathe differently.

It can describe embodiment.

It does not have a body.

That doesn’t make AI useless.

It simply means we should be honest about what kind of intelligence we’re working with.

I Am Not Interested in Purity

I also want to say this because conversations about AI can become strangely moralistic.

I am not interested in proving that my creative process is pure.

Sometimes the answer arrives because I ran a careful comparison across three sources.

Sometimes it arrives while I am walking outside.

I don’t see those things as enemies.

I think the work is learning which kind of knowing belongs where.

AI Can Help Me See the Pattern

One place AI has become genuinely valuable in my work is pattern recognition.

I think in systems.

I collect information.

I notice relationships between things.

I can have fifteen documents, four conceptual threads, a pile of notes, a transit timeline, and a half-developed framework sitting in front of me while some part of my brain insists:

There is a structure here.

Find it.

AI can help me hold more of that material in view.

It can help cluster ideas.

Compare terminology.

Reveal repetition.

Notice where one section contradicts another.

Ask questions I haven’t thought to ask.

That can be incredibly useful.

But recognizing a pattern and deciding what the pattern means are different acts.

That second act still requires interpretation.

And interpretation is where worldview enters.

Experience enters.

Ethics enter.

Context enters.

The human returns to the center.

Sometimes AI Is Wrong. Sometimes I Am Wrong.

This may be the least glamorous part of the whole conversation.

Neither the tool nor the person using it is infallible.

I can misunderstand something.

Misremember something.

Read too quickly.

Get attached to an interpretation.

Miss context.

AI can fabricate a source.

Combine unrelated facts.

Miscalculate.

Give a beautiful explanation of something that does not actually exist.

A responsible process needs room for both possibilities.

This is why verification matters.

And humility.

Not the performative kind where we pretend we know nothing.

The practical kind that says:

I may need to check this.

I think that phrase is going to become more important, not less.

Spiritual Work Especially Needs Discernment

There is an additional layer when we are talking about astrology, intuition, ritual, energy, or spiritual development.

People often come to these practices when they are vulnerable.

They’re grieving.

Changing careers.

Leaving relationships.

Questioning identity.

Searching for meaning.

Trying to understand why something keeps happening.

That creates responsibility.

I never want technology—or astrology, for that matter—to become an authority that overrides someone’s own agency.

AI should not decide someone’s fate.

Neither should a birth chart.

Neither should a tarot card.

Neither should I.

The work should create more awareness.

More language.

More possibility.

More capacity to make conscious choices.

Not dependence on whoever appears to have the answers.

There Are Things I Will Not Outsource

The more useful AI becomes, the clearer this becomes for me.

I am happy to outsource certain kinds of labor.

Sorting.

Formatting.

Initial comparison.

Organization.

Tedious cleanup.

The seventh attempt at getting a meta description under the character limit without making it sound like it was written by a toaster.

Bless.

But there are things I do not want to outsource.

Discernment.

Ethics.

Final interpretation.

Responsibility.

My relationship with a client.

My relationship with my own intuition.

The decision about what belongs in my work.

The decision about what does not.

The part where I sit with something long enough to discover that my first interpretation was incomplete.

The part where lived experience changes the theory.

Those things are not inefficiencies.

They are the work.

What I Want AI to Become in My Practice

I don’t have a completely finished answer.

I suspect none of us do.

This technology is changing too quickly for certainty to be particularly useful.

But I know the relationship I am trying to build.

I want AI to help me become more rigorous, not less.

More curious.

More organized.

More capable of examining a larger body of information.

More willing to compare.

More aware of contradictions.

More able to spend my actual human attention on the parts of the work that require a human being.

Somewhere along the way, my role changed too.

I feel less like the person who has to personally execute every single step and more like the CEO, creative director, researcher-in-chief, and publishing editor of my own work.

I decide what we’re building.

I decide what deserves deeper research.

I decide what fits the larger body of work and what gets cut.

I set the standards.

I review.

I correct.

I send things back when they’re not ready.

And I make the final decision about what leaves the workshop with my name on it.

Oddly enough, learning to work with AI hasn’t made me feel less involved in my work.

It has made me more aware of which parts actually require me.

And there is another part of this that I don’t think we talk about enough.

Having help has given me capacity.

Sometimes that capacity means I can take on a project I wouldn’t have had room for before.

Sometimes it means I can follow another curiosity thread without completely abandoning the first six.

And sometimes it means I can stop.

I can rest for a little while without that anxious feeling that if I step away, I’ll lose the thread of where I was, what I was researching, or what needed to happen next.

That matters to me.

We tend to talk about assistants as though they’re something you’re supposed to earn after you’ve become sufficiently successful.

But who decided you need to be a millionaire before you’re allowed help?

For a solo business owner, creator, researcher, or person building something with more ideas than available hours, AI can function as a kind of support staff—not the person in charge, but something capable of holding notes, organizing information, tracking threads, handling repetitive work, and helping us return to where we left off.

I don’t think freeing human capacity is cheating.

Sometimes it’s how we get enough room to do the deeply human part well.

And sometimes it’s how we get enough room to take a nap.

I support both.

I want the tool to handle more of what is mechanical so I can give more of myself to what is meaningful.

That, to me, is the promise.

Not replacing thought.

Creating more room for it.

And I am beginning to understand that room does not always have to be filled with more work.

The Workshop and the Architect

I keep returning to the workshop metaphor because it feels right.

A workshop contains tools.

Some are simple.

Some are powerful.

Some allow you to do things that would have been nearly impossible by hand.

You learn what each one is good at.

You learn what it is terrible at.

You learn when to use it.

You learn when to put it down.

But the tools don’t decide what you’re building.

They don’t choose the values that shape the structure.

They don’t know why this particular thing deserves to exist.

They don’t carry the responsibility for whether it is sound.

That belongs to the architect.

And in my work, that role remains human.

I use AI.

I value it.

I am fascinated by it.

I expect my relationship with it will continue evolving.

But the research still needs discernment.

The astrology still needs verification.

The symbolism still needs interpretation.

The ritual still needs intention.

The writing still needs a voice.

And ultimately, the work still needs a person willing to stand behind it.

AI is part of my workshop.

It isn’t the architect.

I think I want to keep it that way.

Still curious. Still checking my sources.
Joeii Starr
The Charmed Compass