S

Who is a designer?

I recently read Michael Rock’s 1996 essay Designer as Author for the first time. Its core questions challenge designers to think less about their role title, and more about how their work participates in meaning-making. With AI tools generating images, interfaces, and even strategy, and design engineering blurring the line between design, code, and systems thinking, the definition of a “designer” is more fluid than ever. Rock’s essay feels like a good framework for asking the right questions all over again.

I’ve been thinking about these questions of what makes a designer a lot. As someone who graduated a year ago into the quickly shifting world of product design, at times it feels like I’ve landed into a career path going through an identity crisis. So in trying to find my footing, it’s been grounding to see how relevant Rock’s models of design still feel today.

Even as the materials, tools, and contexts of design have shifted, the models of designer as translator and designer as director feel closely aligned with the kind of work I find myself doing now. I’d even include a new model that was probably there all along: designer as humanist.

Designer as translator of machines and models

Rock’s translator model frames the designer as someone who “remolds the raw material of given content, rendering it legible to a new audience.” As a designer working on AI systems, translation is a big part of the job.

LLMs are the raw material, including:

• The fundamental capabilities of the model (language generation, reasoning, knowledge representation)
• The inherent limitations (knowledge cutoffs, reasoning gaps, ethical guardrails)
• The “language” of prompts, parameters, and system messages the model understands
• The patterns of strengths and weaknesses specific to each model architecture

Translating these often opaque capabilities and constraints involves:

• Recognizing what the models can and can’t do, and how they behave in the wild
• Creating abstractions and metaphors to package complex LLM behavior into simple, user-friendly patterns (is this a chat? a command line? a control panel? a tool?)
• Building conventions for interactions that make invisible processes feel visible and trustworthy
• Contextualizing model capabilities within user workflows and mental models

This translation isn’t neutral or scientific. We need to make interpretive choices that reflect product sensibilities and priorities. Sure, the models work straight out of the box. But we need design thought to craft an experience that feels intuitive and aligned with human expectations. Translation helps us to reframe this raw computational output into a system people can understand, trust, and use.

Designer as director of intelligent systems

If translation focuses on making individual AI capabilities accessible, direction focuses on the orchestration of multiple capabilities into coherent experiences. Rock’s director model describes designers who work on large-scale projects, “orchestrating masses of materials to shape meaning, working like a film director, overseeing a script, a series of performances, photographers, artists, and production crews.”

AI systems likewise have a lot of complex moving parts that need to be orchestrated:

• User inputs
• LLM outputs
• APIs
• External data sources
• Real-time context
• User preferences
• System state

Directing how these parts work together involves:

• Deciding what happens in the foreground vs. the background—what should the user see, and what can run invisibly?
• Controlling timing and pacing of interactions—how do you progressively disclose features so they’re discoverable, but not overwhelming?
• Setting the tone of what the experience feels like, and making sure it matches the user’s mental model

In my experience designing for agents, a lot of the work is sequencing and framing these interactions. We’re not just staging the interface, but guiding the exchange between a person and an unpredictable collaborator. We decide when the agent should take initiative, when it should step back, and how to gracefully shift between the two. This is more than just technical architecture—it’s narrative and rhythm.

A new model: designer as humanist

To Rock’s models of translator and director, I’d add a third that feels increasingly essential when working with AI: designer as humanist. As model capabilities expand, there’s a genuine risk of prioritizing what’s technically possible over what’s humanly desirable. The humanist designer advocates for personal agency and expression in systems that might otherwise prioritize efficiency, scale, or technical performance.

This isn’t about designers all becoming Luddites. It’s just about being intentional. It’s about asking harder questions that go beyond usability: Does this system enhance human capability or replace human judgment? Does it create appropriate transparency or obscure important processes? Does it serve diverse needs or optimize for narrow use cases? Does it foster genuine connection or merely simulate it?

Technologies are never neutral; they reflect choices and values, and shape experience and behavior. This humanist lens isn’t separate from design work—it’s integral to it, informing decisions about what to emphasize, what to hide, what to automate, and what to leave in human hands.

To conclude: let’s just design

Rather than seeing the rise of AI and design engineering as a displacement of traditional design, we might instead understand it as a (natural) evolution of the relationship between design and implementation—one that calls for closer collaboration, deeper technical fluency, and a shared language across disciplines.

Designers should embrace our role as translators and directors of complex, evolving systems. Design today is less about crafting finished products and more about crafting the conditions for meaningful experiences to emerge—particularly in how people interact and collaborate with AI.

It's a huge challenge to design for agency in a world of automation. The people solving this will need engineering acumen and human empathy, systems thinking and ethical judgment, comfort with uncertainty and commitment to human values.

Luckily, these aren't new jobs! Translating complexity, directing experience, and centering human values is the same crucial work we've always done, just adapted to the times. Perhaps this is the enduring value of the designer—carrying these fundamental practices forward while embracing new possibilities. At least that’s the kind of designer I'm trying to be :-)