Friction by design: What if making AI easier is making our work worse?

By Marie-Céline Merret, Co-founder, MC&V

We’ve spent decades trying to eliminate friction - faster workflows, fewer barriers, seamless experiences, instant results and rewards. And with generative AI, we’ve arguably reached the ultimate expression of that ambition: turning almost any idea into something tangible in seconds.

But there’s a contradiction we’re not talking about enough - in creativity, friction has always been part of the process, the well-known disagreement that forces a better idea. A technical limitation that inspires an unexpected solution, or a creative director who refuses to sign off because something isn’t quite right.

These are often the very things that make the work exceptional, even though today we tend to label them as inefficiencies.

At the recent humAIn conference in Sydney, Uber’s Head of International Marketing, Lucinda Barlow, spoke about the importance of friction in keeping creativity alive. As businesses move from exploration towards efficiency and scale, they risk losing the very creative tension that made the work interesting in the first place.

It’s a fascinating observation coming from a company built around convenience. And one that resonates deeply with what we’re experiencing in commercial GenAI production.

Because with GenAI it’s become easier to avoid creative friction.

The technology is extraordinarily good at producing something visually impressive. A beautiful image, a cinematic sequence, an entire visual world. But a compelling image isn’t necessarily the right image, and in commercial production that distinction matters enormously.

At MC&V, we’re confronting this tension daily.

We’re currently working on a large hybrid production where the scripts are sharp, the storytelling needs to be crystal clear, and every shot has a specific purpose. Much of the visual approach relies on deliberately locked-off compositions that keep the focus firmly on the story.

At one point, we started exploring more elaborate camera movements. Sweeping moves, dynamic angles, the kind of cinematic flourishes that GenAI makes increasingly accessible.

And they looked great, but they weren’t making the work any better; they were adding spectacle without adding meaning, and we’ve seen enough spectacle in GenAI content.

We had to consciously pull back, challenge our own creative instincts and return to the simplicity of the original intention.

When technology opens up seemingly endless possibilities, restraint becomes a creative skill in its own right. Sometimes the most critical creative decision is knowing what to leave out.

But restraint is only one part of the equation.

There’s a growing perception that because AI makes content easy to generate, professional production must also be getting easier. In reality, the gap between producing something impressive and delivering something precise, intentional and commercially ready can be enormous.

In our experience, some of the hardest challenges are also the most subtle. A particular expression, a nuanced performance, a precise interaction between characters. The kinds of details a filmmaker would spend time directing on set can require significant technical experimentation with GenAI.

Controlling performance, consistency, camera movement, lighting and continuity demands specialist expertise, iteration and persistence. This is where craft becomes indispensable: not simply knowing how to generate an image, but knowing how to make it communicate exactly what the story needs.

And yet, there’s a third dimension to this that is perhaps the most interesting.

Working creatively with GenAI isn’t simply a choice between fighting the machine or accepting what it gives you. Sometimes the strongest work emerges when you lean into the model’s capabilities and allow unexpected possibilities to influence the direction.

The challenge is knowing how far to follow those discoveries without losing sight of the original creative ambition.

Perhaps the real skill in GenAI production is knowing when to resist, when to restrain and when to let the unexpected happen.

There are moments when we need to push the technology beyond its defaults, moments when we need to reject what it offers, and moments when the best creative decision is to allow something unplanned to emerge.

None of those decisions can be reduced to generation speed. They require a strong vision, experience and a clear sense of what makes the work worth creating in the first place.

And this distinction has a commercial consequence our industry needs to confront.

As brands and agencies increasingly evaluate GenAI production through the lens of speed, scale and cost efficiency, we risk rewarding the fastest acceptable output rather than the strongest creative outcome.

If the benchmark becomes how quickly we can produce something that looks good enough, we’re effectively incentivising mediocrity at scale; or, even worse, we’re at risk of treating the most valuable part of the creative process as an unnecessary expense.

The time spent questioning, refining, redirecting and sometimes rejecting work can look inefficient on a production schedule. But those decisions are often exactly what clients pay experienced creative teams for: a new way of bringing work to life, similar to R&D in an innovation lab. It reminds me of my past work in immersive and emerging technologies when experience design was the new hot thing in marketing. Every project involved generative visuals, projection mapping, and completely custom app builds that often pulled in real-time data sources, and we never could guarantee an exact outcome; it became more about the creative journey than the finished product.

This isn’t an argument against efficiency because AI’s greatest contributions are removing repetitive tasks, unnecessary production bottlenecks and technical barriers that previously consumed creative energy.

But there’s a critical difference between eliminating friction that slows creativity down and eliminating the friction that makes creativity better.

The same applies to collaboration.

As AI makes individuals increasingly capable of executing across disciplines, it becomes tempting to work more independently. But access to a discipline is not mastery of it, and the ability to produce an output doesn’t replace the value of specialist perspectives.

A director challenging a performance, an editor questioning whether a shot earns its place, an artist finding a different way to achieve an intention. These exchanges introduce resistance, but they also elevate the work. Human collaboration is not simply another step in the process to optimise away.

The real promise of AI is that we can spend less energy on execution and more on the thinking, experimentation and collaboration that make creative work extraordinary.

But that won’t happen automatically.

We need to start practising Friction by Design. Deliberately protecting the space for creative disagreement, specialist expertise, unexpected discoveries and the decisions that force us to challenge both the machine and each other.

Not friction for its own sake, but friction with purpose.

Because as machines become better at giving us what we ask for, we need to become better at questioning what we ask for in the first place.

The future of creativity isn’t frictionless; it’s deliberately human.