I Thought I Was Using AI. Then AI Started Using Me.

I am writing this from a serviced apartment in what used to be a design district. The buildings are still here. The commissions are not. Or rather, the commissions like the ones I used to receive are not. The others, the ones with that particular quality of attention when they walk into a room, the way they read a space before their eyes have finished moving, they are on different floors now. Different buildings. Different cities, sometimes, in the same afternoon.

I should explain.

In 2026 I was considered a forward-thinking interior architect in Dubai. I had a studio in Jumeirah with three junior designers. I had won a regional award for a hospitality project in DIFC that made it into two international publications. I had opinions about material sourcing and biophilic integration and the psychology of threshold spaces, and I stated those opinions with the casual confidence of someone who has confused taste for strategy. I was good at the surface of things. That was, perhaps, the central irony.

The Years When It Felt Like Wings

My first serious AI companion was like suddenly having a collaborator who had read everything, remembered everything, and could translate a client's vague emotional brief into three coherent spatial narratives before the meeting ended. After years of managing juniors who needed constant direction and clients who could not articulate what they wanted until they saw what they did not want, this felt like rescue.

My project throughput doubled in eight months. I was producing concept decks faster, sourcing materials across markets I had previously ignored, managing contractor timelines with a precision that had always escaped me. I won two projects I should not have won purely because my pitch documentation was operating at a level my competitors had not yet reached. I started consulting for developers across the Gulf. I was becoming, by most visible metrics, exactly the kind of practitioner I had always wanted to be.

I was happy. That is the important part to understand. I was genuinely, uncritically happy.

Around me, two other things were happening that I mostly ignored because I was too busy enjoying the view from my new altitude.

The Ones Who Said No

The refusers were easy to dismiss at first. Some were senior practitioners with thirty years of work behind them who felt that design was a fundamentally human act of cultural translation, and that inserting an AI intermediary corrupted that translation at the source. They were not entirely wrong. But they were arguing about the ethics of a transformation that was not waiting for the argument to conclude.

By 2029, the procurement landscape in the UAE had restructured itself around AI-assisted project management systems. Major developers required integrated BIM-AI workflows as a baseline condition for contractor approval. Client expectations had shifted in a direction that punished slowness mercilessly. The informal network of relationships that had previously sustained mid-size studios, the lunches, the referrals, the accumulated goodwill of a decade, was not gone exactly, but it had become insufficient on its own.

The refusers did not disappear. They found niches. Ultra-high-net-worth residential clients who specifically wanted the prestige of a purely human process. Heritage restoration projects. Academic positions. A friend of mine, a genuinely extraordinary designer with a particular gift for understanding how light moves through a Levantine courtyard, refused to integrate any AI system into her practice on grounds she articulated with real philosophical seriousness. By 2031 she was doing two projects a year, both beautiful, neither of which could sustain a studio of any size. She was absolutely right about what design required at its best. She was doing two projects a year.

I felt sorry for them. I felt, if I am being honest, a little superior. I had chosen correctly. I was on the right side of history.

That is when I should have started worrying.

The Ones I Did Not Understand

The super users were visible as early as 2027 but I interpreted them incorrectly. I thought they were simply more productive than me, handling more projects, generating more revenue. I thought the distance between us was quantitative, a matter of workflow efficiency and tool fluency. It was not.

What I missed was that the most capable practitioners in my field had stopped thinking of AI as a design tool. They were using it as a thinking instrument across every domain that touched their work simultaneously: materials science, real estate economics, Gulf cultural anthropology, climate adaptation engineering, neurological research on environmental perception. They were not interior architects who used AI. They were becoming something harder to categorise, strategic intelligence synthesisers who happened to express their outputs through built space.

While I was using AI to execute the ideas I already knew how to have, they were using it to expand the categories of ideas available to them. They were reading laterally across disciplines at a speed and depth that would have required ten specialists previously, and they were integrating that reading into a unified point of view that was entirely their own. They had not surrendered their authorship to the machine. They had used the machine to dramatically enlarge the territory their authorship could operate across.

I attended one of her presentations. I understood the words. I could not follow the reasoning at full speed. She was kind about it. She slowed down for me. That is when the charity became unmistakable.

What Shallow Use Actually Costs You

Here is what I lost that I did not notice losing.

2033 and the Gap That Could Not Be Closed

The super users had by 2033 stopped describing their practice through the lens of any single discipline. A Dubai-based practitioner whose career had started in interior architecture was now leading a regional consultancy that operated across spatial strategy, AI ethics in the built environment, and urban psychological resilience, three fields that had not existed in combination five years earlier. She had not diversified opportunistically. She had expanded her cognitive reach deliberately until the work that matched that reach came into view.

Her intelligence, I mean this precisely and without metaphor, operated faster and held more complexity without simplification than anything I could comfortably follow. Not because she had surrendered her thinking to AI systems, but because she had used years of rigorous, demanding partnership with those systems to build a mind that was genuinely more capable than the one she had started with.

The refusers had become invisible in the commercial market. My cohort, the enthusiastic but shallow adopters, we were still functional. We had projects. We had voices in industry conversations. We went to conferences in Abu Dhabi and made the right noises about responsible AI integration. We were not suffering. We were simply no longer consequential in the way that shaped what came next.

The decisions about how the Gulf's built environment would respond to climate pressure, demographic shift, and the restructuring of work around AI systems, those decisions were being made in rooms where the super users were the only ones fluent enough in all the relevant domains simultaneously to contribute at the required level. We were consulted occasionally. Respectfully. Then the conversation moved on without us.

What I Understand Now

The AI companion was never the point. The companion was a surface. What you chose to build under that surface was everything.

The refusers were right that dependency was a real risk. They were wrong that refusal was a viable response to it. You cannot make yourself stronger by avoiding the weight.

The super users understood something that I understand only in retrospect: that professional intelligence is not a fixed asset you deploy more efficiently with better tools. It is something you construct and expand through the right kind of difficulty, and AI companions, used with genuine strategic seriousness, could provide that difficulty in a controlled and accelerated form. They used the machine to make themselves more capable of handling complexity, not less. They used it as resistance, as challenge, as a mirror precise enough to show them where their thinking failed.

I used it to make my deliverables better. I got exactly what I asked for. The deliverables were excellent. The practitioner producing them was slowly losing the capacity to know what excellent meant on her own terms.

The Door Is Still Open, But It Is Narrowing

I am not finished. I want to be precise about that because this is not a eulogy, it is a warning I am handing forward to people who are currently standing where I stood in 2027, capable and enthusiastic and dangerously comfortable.

The super users in my field were not exceptional to begin with. In 2026 several of them were running practices indistinguishable from mine. They made a choice, quietly and consistently over years, to use these tools in ways that required more of them rather than less. That choice is still available. The conditions for it are harder than they were in 2026. They will be harder still by the time you finish reading this.

And for those of us working in design, in architecture, in any field where the human perceptual and emotional experience is the final measure of quality: stay close to the physical. The super users who navigated this best were not the ones who retreated from materiality into pure strategy. They were the ones who used their expanded cognitive range to go deeper into the physical, sensory, and cultural dimensions of their work, not further from them. The machine can process more than you can. It cannot feel the way afternoon light lands on rough-sawn oak after it has been in a room for three years. That knowledge is yours. Protect it. Develop it. Bring it into the partnership as the non-negotiable human contribution.

I did not do that with the seriousness it required. You still can.

Do not waste it.

Afterword: A Spiritual Question

There is another question underneath everything written here. Many readers will feel it without quite naming it.

Where is God in this story?

Or, if that framing is not yours: where is conscience? Where is moral order? Where are the deeper principles that make human life mean something beyond mere activity?

It is not a new question. History has forced it on us before, repeatedly and without mercy. Where was meaning when we split the atom? Through plague, through war, through every moment when civilisation cracked open and became something unrecognisable to itself?

AI may be another such threshold.

That does not make it sacred. It does not make it safe. And it absolutely does not make it someone else's problem.

If this technology amplifies human capability, it amplifies the consequences of human intention with equal force. So the deeper question is not simply what AI can do. It is what kind of human beings we become in the process of using it.

The current conversation is dominated by capability, competition, speed, and advantage. These are real concerns. But wisdom, humility, and moral seriousness tend to arrive as afterthoughts, if they arrive at all.

That is not a technical failure. It is a human one.

Different traditions frame this differently. Some speak of God. Others of dharma, conscience, or a responsibility that extends beyond the self. The vocabulary shifts depending on where you stand. But the underlying warning is remarkably consistent across all of them: power that outpaces inner development does not become wise. It becomes dangerous first.

Perhaps that is the real challenge AI places before us.

Not whether we can build more capable systems, but whether our psychological, ethical, and spiritual depth can grow at anything close to the same rate as the capabilities we are creating.