Rediscovering sketching in the age of AI

Over the past two years, few technologies have generated as much discussion in automotive design as generative AI. New tools have appeared at a remarkable pace, promising everything from instant concept generation to presentation-ready visualisations. Alongside the excitement has come a persistent question: what does this mean for the role of the designer, and the future of traditional skills such as sketching?

Much of the early discussion around AI focussed on idea generation. Could AI produce new forms, unexpected combinations or concepts that a designer might never have imagined? Others questioned whether designers would still need to sketch at all.

To answer those questions, it first helps to ask a more fundamental one: why do designers sketch in the first place? That's a question designer, educator and researcher Michal Jelinek has spent much of his career exploring. His doctoral research examined sketching not simply as a way of illustrating ideas, but as an integral part of the thinking process itself.

Talking with Car Design News, Jelinek argued that sketching helps designers think by externalising ideas. Rather than trying to hold every possibility in memory, designers can respond to what appears on the page, continually discovering relationships, opportunities and new directions as the sketch develops.

Designers do not always know exactly what they are looking for before they begin sketching. A line may suggest a surface, a change in proportion or an entirely new direction. At the same time, a sketch communicates just enough information – proportion, stance, character and intent – for others to understand the idea, while remaining open enough for it to evolve. The ambiguity in a sketch is part of its value.

Softly sketched lines may suggest several possible surfaces, while an unfinished area invites further interpretation by the designer or colleagues, encouraging discussion and further iteration rather than closing down possibilities.

AI may have revealed something about sketching itself: even the rough sketches of an experienced designer can contain a remarkable amount of design information, ready to be interpreted by designers and modellers, and now by AI

Jelinek argues that sketching is not about creating or producing an artwork. "A sketch is about producing an idea, not an image. It's basically an experiment that leads you from no idea to an idea." Even the repeated sketch lines often seen in designers' drawings are not signs of indecision, but of exploration. As he explains, "We are exploring the space, exploring to find and discover the right curve." The value of sketching therefore lies less in the finished drawing than in the interaction between the designer and the sketch, where ideas are generated, tested and refined.

Rather than disappearing when AI enters the process, that ambiguity becomes something AI can respond to. Royal College of Art graduate Henrik Svanvig has developed a Vizcom workflow that combines rough sketches, reference imagery and collage-like compositions before using AI to rapidly develop those ideas into more complete design proposals. Rather than aiming for a finished rendering, the process becomes a conversation about materials, mood and theme, not just the form. The designer sketches, evaluates the AI's interpretation, paints over it, and introduces new references before generating another iteration. Instead of replacing sketching, AI dramatically shortens the sketch–feedback–sketch loop.

What has become increasingly apparent is that AI can extract far more from a sketch than many designers might have expected. Even a handful of loose lines can convey proportion, stance, character and design intent, while leaving many details deliberately unresolved. AI may have revealed something about sketching itself: even the rough sketches of an experienced designer can contain a remarkable amount of design information, ready to be interpreted by designers and modellers, and now by AI.

Increasingly, designers are using AI less as a creative randomiser and more as an interpreter of the design intent contained within a sketch. Rather than simply generating unexpected images, it can develop a rough sketch into consistent views and progressively refined visualisations, evolving from an experimental curiosity into a practical creative partner.

In conversation with Course Director Shaun Hutchinson on the Coventry Connections podcast, Dongfeng Exterior Design Manager Jamie Barrett described how AI is allowing designers in large studios to create their own visualisations and animations much earlier in the design process, without relying on specialist visualisation teams.

An early sketch can quickly become multiple consistent views of the same proposal – front, side and rear – without first building a complete 3D model. Designers can carry ideas much further themselves, exploring and communicating concepts more quickly. The result is not simply faster visualisation, but a shift in how designers spend their time. Designers can spend less time producing presentation images and more time exploring and evaluating what they have designed, bringing a broader range of developed proposals into design reviews. As Barrett puts it, AI has "really allowed us to focus more on design and less on those laborious tasks."

Selecting the right direction, recognising when polished imagery disguises weak design thinking, and refining ideas become increasingly important skills

General Motors recently illustrated a similar shift in an article describing its AI-assisted design workflow. According to designer Daniel Shapiro, a process that traditionally took multiple teams several months – from sketch through 3D model visualisation to animation – can now be completed by a single designer in less than a day. The benefit is not simply speed, but the ability to explore a wider range of ideas before narrowing down to a single direction.

As AI makes it easier to turn rough sketches into convincing proposals in minutes, the designer's judgement becomes more important than ever. Designer and educator Berk Kaplan has observed the same shift in his teaching. AI allows students to generate an abundance of polished concepts in a very short time, but many struggle to evaluate what they have produced. Rather than lacking options, they lack the experience to curate them.

Kaplan recalls one particularly talented student arriving with a board full of impressive AI-generated concept renderings. "The student said to me, 'Look at these results. I don't know how to decide. They’re all good.'" Rather than struggling to generate ideas, the student was overwhelmed by choice.

For Kaplan, AI is not replacing designers, but shifting their value. Selecting the right direction, recognising when polished imagery disguises weak design thinking, and refining ideas become increasingly important skills. The challenge is no longer producing enough ideas, but developing the experience and design intent to recognise the strongest solution.

Henrik Svanvig also cautions that AI works best when it is guided by clear design intent. In his own workflow, the greatest value from AI comes not from defining the overall volume of a vehicle, but from rapidly exploring unresolved areas and technical details. "I've had the greatest results with AI when I've had a clear design intent," he says, while warning that inexperienced designers can become distracted by polished imagery rather than the underlying quality of the design.

If AI increasingly takes over the work of visualising ideas, the designer's value shifts further towards exploration, judgement and decision-making

If professional workflows are changing this quickly, design education cannot stand still either. Eric Stoddard, founder of Car Design Academy, says the school is reconsidering parts of its curriculum in response to what is happening inside studios. If an early sketch can quickly become a rendered image, several views or even a short video, it becomes reasonable to ask how much time students should spend learning traditional presentation techniques.

The teaching emphasis may shift towards the parts of the process that AI cannot easily supply: creativity, proportion, packaging, visual judgement and the ability to develop a strong original idea. As Stoddard put it during our conversation, "sketching remains the spark" – the starting point for ideas that AI can rapidly develop and illustrate. He also argues that "AI doesn't replace designers. It amplifies the good ones. But only if the designer has a clear process."

For an industry understandably fascinated by image generation, this may be the more important story. If AI increasingly takes over the work of visualising ideas, the designer's value shifts further towards exploration, judgement and decision-making.

AI has not made sketching obsolete. It has revealed why designers sketch in the first place. The designers getting the most from AI are those who use it to explore, refine, challenge and accelerate the development of their ideas. The designer’s sketch remains where those ideas are first discovered.