Can artificial intelligence replace (graphic) designers?

In a post-screening discussion of Graphic Means last month at HKDI, a member of the audience posed an interesting question: ‘Can artificial intelligence replace designers?’ Online tools such as GraphicSprings can already generate logos based on a series of criteria and preferences, and can apply the logo on various items with rather convincing visualisations. But is this really design?

Many years ago I bought a secondhand copy of Ken Garland’s small book on graphic design humbly titled Graphics handbook, published in 1966 (with a striking cover, by the way). In the most unlikely of places — a small instructional volume on the professional practice of graphic design — are some of the most progressive and insightful predictions of what we designers are now facing.

In the introduction (set in an unusually large type size to highlight its importance) are some very important words of wisdom:

Why, at a time when communication systems of all kinds are increasing in scope and complexity, we should feel bound by any narrow definition of the scope of graphic design, I don’t know. But we still are.

Garland went on to describe that perhaps a properly instructed computer will be able to generate layouts, and the designer will then be devoted to the creative task of programming the computer with appropriate parameters. I know for a fact that Buro Petr van Blokland + Claudia Mens was already doing this 16+ years ago [unfortunately their old website no longer exists].

As a rule, what can be automated will be automated, and it’s usually the drudgery work that gets automated first. In the case of graphic design, that means the production of graphics (and the churning of quantitative data). I’m no expert in AI technologies, but perhaps it’s unlikely that AI can generate things that no one has seen before (ie to innovate). This is what designers do in order to push creative boundaries and to differentiate from others. Another task that is unlikely to be replacable is user research — to understand the needs of users and contexts associated with a design problem and generate appropriate solutions. People’s behaviours are extremely nuanced and complex, and not strictly quantitative. Contextual inquiries for design are most likely to be qualitative, and the analysis and synthesis of such nuanced qualitative data is probably not yet possible with AI. And design solutions are not straight-forward answers. Many equally valid solutions are often possible, and the move from research insights to final solution often requires creative leaps rather than linear calculations.

Garland closes the introduction by saying:

Only two things could stop the graphic designer from growing up with the rest of the technological world: a failure to familiarise himself with the new areas of knowledge springing up alongside his own; and an inability to free himself from the strait-jacket of his preconceptions about the kind of tasks with which he expects to be confronted.

I couldn’t agree more, especially when we now live in a highly complex, technologically-mediated world. To think that (graphic) designers are only concerned with the visual couldn’t be more unwise.