Berk İlhan

Executive Summary

Berk İlhan doesn't see AI as artificial intelligence. He sees it as an extension of human intelligence – a human-made tool that co-evolves with creative practice rather than replacing it. His work spans product design, interactive installations, and immersive experiences, rooted in IDEO-influenced design thinking and a conviction that inspiration is the essential nutrition of creativity.

His AI integration began in 2022 with Midjourney, where he immediately recognised the technology's power to accelerate the translation of ideas into communicable visual form. He has since settled on ChatGPT as his primary platform, valuing its conversational iteration over prompt-revision models. He resists switching to newer tools – not out of inertia, but because he has built a working relationship with the system and values what that continuity provides. He has invested time learning how it interprets particular words and phrases, developing a feel for its sensitivities in the same way one learns to work with a human collaborator. You don't swap collaborators lightly, and he applies the same logic here.

I already learned it and we invested in the collaboration and there's so much data — it knows me.

The relationship has become genuinely reciprocal in his view, though he maintains clear authorship boundaries. Rather than raw material alone, AI functions as a powerful synthesizer and rendering partner – actively interpreting, recombining, and proposing, while Berk's creative vision acts as the compass. His prompts carry intention but also evolve through interaction, making the process less about shaping passive material and more about a dialogue where he guides direction through intention. AI operates as both synthesizer and medium.

His most honest observation is also his most useful: AI can disrupt creative flow as easily as it enables it. The speed of generation can eliminate the gradual discovery process that builds emotional investment, and the sheer volume of output can overwhelm rather than inspire. Beyond volume, there is also the challenge of unpredictability – unlike traditional tools that become extensions of the hand through mastery, AI's underlying system keeps evolving even when the interface stays the same, creating a moving target.

Sometimes those conversations with AI take so much brain energy... by the end of that conversation, I don't feel like I have energy left to actually design the thing.

Managing this tension – between acceleration and depth, between AI's generative abundance and the designer's need for focused intention – defines his practice.

Visualisation and the acceleration of iteration

Berk's most consistent use of AI is in early-phase concept visualisation – creating imagery that communicates ideas to stakeholders before detailed design work begins. For a science centre project in 2022, when AI tools were still in relatively early phases, AI-generated visuals made abstract concepts tangible before the project was fully defined, producing imagery that felt almost realistic but retained a dreamlike quality. This situates the experience historically: since then, both the tools and Berk's prompting approach have evolved significantly, allowing for much more precision and intentionality.

This proved double-edged. Stakeholders began responding to incidental details that AI had generated without intention – elements that were visually compelling but irrelevant to the core concept. His solution was to present a range of visions explicitly framed as directional rather than definitive, making clear that the work was exploratory. Managing expectations around AI-generated imagery became its own design challenge.

For the Smile Mirror – an interactive mirror project that began in 2014, long before AI tools existed – the work was developed through research, analog and digital sketching, CAD, prototyping, coding, sensors, and hardware production. AI came much later as a support tool for visualisation and exploration, not as part of the core creation. This distinction matters: the project predates AI-driven design entirely, and should not be read as an example of it. The piece – where the mirror only reflects when the viewer smiles, inspired by research with a cancer survivor who struggled with her self-image during diagnosis – used AI-generated renderings with specific material and lighting qualities to explore and communicate the concept. The final design still required simplification when production reality couldn't match the AI-generated vision.

This gap between AI's visual possibilities and the constraints of making is a recurring theme. AI generates details that are impossible to recreate, creating distance between the excitement of visualisation and what can actually be built.

The collaboration investment: learning AI's sensitivities

Berk describes his relationship with ChatGPT in terms that closely mirror how he talks about human creative partnerships. He has spent time learning how the system interprets particular words and phrases – noticing, over repeated interactions, that it understands certain terms in specific ways – and resists switching to newer tools because that accumulated understanding would be lost. Platform loyalty, for him, is the rational behaviour of someone who has built a working relationship and values the continuity it provides.

He personifies the system in small but telling ways – noticing something alive in the text generation process, finding that the interaction design of ChatGPT, especially the micro-vibrations and the way responses appear, creates a visceral, almost life-emulating experience. This keeps the sense of "aliveness" grounded in design rather than speculation, and reflects genuine attentiveness to the system's qualities and tendencies.

Authorship, control, and the fabric metaphor

Berk's authorship position is clear and consistently held. AI functions as both synthesizer and medium – actively interpreting and proposing – while the designer brings intentionality, aesthetic judgment, and final decision-making to what it generates. Full control, even without AI, is rarely possible: outcomes are always shaped by materials, collaborators, constraints, and production realities. Trying to fully control the result can sometimes work against the natural evolution of the process. The more accurate framing is that the work should be guided by intentionality, aesthetic judgment, and decision-making rather than absolute control.

This reproducibility standard matters practically. For cabinet designs shared on Instagram as process work, AI-generated imagery is acceptable. For final products delivered to clients, the design must be consistent across iterations – if it changes with each generation, it hasn't yet become a design. That said, AI tools have evolved significantly, and a level of precision and consistency that was not possible in earlier stages is now achievable. The distinction between sketch and final is becoming more fluid. What remains constant is the standard: if the outcome reflects Berk's creative vision, vocabulary, and aesthetic judgment, it can be used – regardless of whether it was produced through AI or traditional methods.

He is sceptical of vague prompting, preferring precision. He rarely asks open-ended questions that produce unpredictable results, because he wants to be able to claim the output as genuinely his own – grounded in his imagination and intention rather than the system's tendencies. The concern isn't that AI lacks capability. It's that without precision, the work risks belonging to the algorithm rather than to him.

Speed, flow, and the lost joy of discovery

Berk's most candid observation about AI concerns what it takes away as much as what it provides. Drawing on Csikszentmihalyi's concept of flow, he identifies two failure modes: AI providing too much information (overwhelming) or too easy solutions (boring), both disrupting the optimal creative state. The gradual discovery process – the one that builds emotional investment and playful problem-solving – gets bypassed when AI generates answers too quickly. It takes the fun out of it.

Beyond volume, there is a further dimension: unpredictability. With traditional tools – sketching, CAD, rendering – you gradually master the system and it becomes an extension of your hand. With AI, even when the interface stays the same, the underlying system keeps evolving. That creates a moving target. Sometimes the disruption is not just the volume of output, but the friction of trying to control or refine something that doesn't behave consistently – fixing a tiny detail, for instance, can take significantly longer than generating the first set of images.

He also notices design agencies using AI to jump directly to solution iteration rather than ideation, skipping the exploratory phases that give design work its depth. His own formal process – the double diamond, structured timelines, ideation phases – has become harder to maintain. The tools pull toward output; the methodology requires resistance.

What he envisions as genuinely useful is AI that helps maintain process discipline rather than bypassing it – a system that notices when the designer has jumped ahead and brings them back to the right stage.

Hey, you're going too far out of process. Come back to the process. This is what we should do next.

Not an output generator, but a process partner.

Interface vision: the studio as living room of ideas

Berk's frustration with current AI interfaces is specific and revealing. The list-based conversation archive – a chronological stack of different creative activities – fails to reflect how creative work actually organises itself. A mood board session, a CV draft, and a product ideation conversation appear identical in a text-based list despite serving entirely different cognitive purposes.

What he envisions is spatial:

a three-dimensional main space that you create like... the studio, this is the living room of my ideas.

Different zones for different cognitive modes – a large table for sketch papers and ideation, a pin board for mood boards, a dedicated space for speculative thinking – with physical metaphors triggering different kinds of thought. The underlying conviction is that environments shape cognition, and tools that ignore this miss an opportunity to genuinely support creative practice rather than merely accelerating output.