When everyone can create content, how do we create quality?

With AI, making a film is easier than ever. But how do we make it good?

Anyone can make a film with AI. But can anyone make a good film with AI? For those of us involved in commissioning and the production of visual content, that’s becoming an increasingly important distinction.

When working with film teams to create content for events, exhibitions and experiences, it’s easy to become focused on the shiny, alluring tech itself. What can AI do? How quickly can it do it? What might it save us?

But perhaps the more useful question is: how do we use AI well? How do we make standout, impactful work with it?

With a background in scriptwriting across film, television and radio, I’m fascinated by the possibilities AI has opened up. But I’m less interested in the hype than the practical, pragmatic reality.

Where does AI genuinely add value? Where does it fall short? And what do the rest of us need to understand if we’re going to help our film teams create the best possible work?

To find out, I asked the people using these tools every day: our talented, relentlessly-curious film team.

Where AI adds value (and doesn’t)

AI tools can explore possibilities that sit beyond traditional production constraints. Concepts can be tested, challenged and refined more quickly. Ideas can become visualized from words in seconds, moodboards to moving images, storyboards to living breathing scenes.

Need to visualise a metaphor that would be impossible to shoot? Show an environment that doesn’t exist? Build animatics or proof-of-concept films that allow stakeholders to react to something tangible before deciding whether it’s worth building for real?

Yep, AI can help with that.

One of our film team said that they’re now able to pitch ‘films’ instead of ‘videos’ – AI enabling us to sell ideas that were too outlandish, too metaphorical, or simply too ‘expensive’ even just a few months ago.

It can also help us get closer to an idea, faster. If we’re doing it right, and protect the time to do so, we can spend more where it might matter most – interrogating an idea or concept.

There’s an argument too that – in some cases – AI tools are making films feel more authentic. Documentary makers are re-skinning the voices of vulnerable participants to anonymize their contributions but retaining real emotion. You can now hear your favourite stars sound like themselves in another language – rather than being dubbed by a different actor entirely.

Deeper into production, labour-intensive processes can now be completed in a fraction of the time. Motion capture workflows in minutes instead of days. Content revisited and reworked without fear of derailing a project.

But, we have to be honest about the inflexibility of generated video – it’s still not that easy to ‘tweak’ images and change elements baked into a shot without regenerating it wholesale.

Regeneration also risks further continuity problems. You never quite know what else will change in your frames when you roll the dice. But if your project, cutting and story-telling can plan for this inflexibility – and you’re prepared for this shortcoming – it’ll help.

The need for planning, for clarity of brief, concept and direction remains.

At its best, AI in film doesn’t necessarily replace human creativity. It accelerates the laborious bits in-between the human choices – short-cutting the hunt for stock footage, slashing render times, shaping better, tighter briefs up front – so we can get to the valuable creative decisions more quickly.

The need for human discernment

AI can generate an endless stream of options. Images, scenes, scripts, voices. But, someone has to decide what’s worth keeping (and losing) along the way. Taste – and the ability to recognise what belongs and what doesn’t – becomes more valuable in a world of abundance and infinite possibility, not less.

The same applies to originality. AI can explore, combine and iterate on existing ideas. But the leap to something genuinely original – an idea rooted in a fresh perspective, an unexpected insight or a uniquely human experience – still belongs to us. The role of AI shouldn’t be to replace original thinking. But to give it more room to flourish.

The cinematic grammar that guided filmmakers a century ago still matters. Prompts are words – so an understanding of filmic terminology will get us close to what we want to achieve. Composition, pacing, tension, performance and storytelling still matter – as do the gut choices that can be made after honing such skills for years. The tools may be changing, but the fundamentals remain remarkably consistent.

Importantly, AI doesn’t know why a film should exist in the first place. It doesn’t implicitly understand what an audience needs to hear – or why one idea might emotionally resonate more deeply than another. It cannot (yet!) discern in the way human talent can.

And, although AI is becoming better at imitating human expression and movement, audiences have always been remarkably sensitive to lack of visual authenticity. The ‘ick’ is real – that all-too visceral bristle we feel when we’re being asked to assume an image is real when we know it isn’t. The slick, gloopy sheen of AI is something we’re (still) having to work around and mitigate for.

Navigating the (current) limitations of generative AI in film – the ick, inflexibility of image – is a constant, evolving discussion amongst those playing with the tools first-hand. Although capabilities are continually evolving, we need to be honest with colleagues and clients about what AI can – and can’t – contribute to a project.

Technology can help us make films in ways it couldn’t before. But humans are still what makes them worth watching.

Starting with the right questions

Spoiler. ‘Can we do it in AI?’ is unequivocally not the right question to start with.

And yet, it is the one we predictably hear more and more at the off. From clients, from team members, from those with roles interested in primarily finding efficiency and value – and maybe, most shockingly of all – from our own mouths. (Yes, I mean you at the back).

But if we want to ensure lasting impact, quality and authenticity, there’s surely better things to ask ourselves as we kick off a film project:

What are we trying to achieve? What do we want to say? What does our audience need to hear? What impact or emotional response are we trying to invoke? These starters-for-ten remain as relevant as they ever were. And, if we do have to bring AI up at the off, maybe let’s frame it like this – Where can AI genuinely help this project be better?

As ever, the most effective briefs don’t prescribe the solution. They define the challenge.

These questions matter even more when AI becomes part of the process, not less. The clearer the objective, the easier it is to decide where AI might help.

Is speed the priority? Is there a need to visualise something quickly? Create something that would be difficult, expensive, or even impossible to produce traditionally? Are we aware of the challenges in choosing this mode of production?

Equally important is understanding what must feel authentic. Where does the emotional connection come from? Which parts of the story demand a human touch, real performance or carefully crafted creative judgement?

So, bring the why. Be clear about the outcome you’re trying to achieve. Then allow the filmmakers, creatives and technologists to help define the best route to get there.

Surrendering autonomy

We can now arrange AI systems together to produce a complete film, from a single prompt. Now, it might not win awards, but it might be okay. So why not just surrender our autonomy to automation?

The point isn’t that AI should be avoided. It’s that we shouldn’t let it be doing our thinking for us. At least, not the thinking that matters. Creative work still depends on the right person deciding what matters, what feels right, and what should be left out.

In our discussion, there was warning against outsourcing such critical thinking. And, what happens when creative choices are handed over too readily. Because once we stop asking why something should be made, we begin to lose the point of making it in the first place.

AI can help remove friction. It can speed up exploration, reduce labour and open new creative possibilities. But, if that convenience means less discernment, less authorship, then something essential will be lost. And that matters even more in a world where content is easy to generate but much, much harder to distinguish.

Where we don’t make the creative decisions, AI will make them for us. Left unguided, AI models default to patterns in their training data, shaping everything from composition to casting. In a world where anyone, from anywhere, can be generated and dropped into a scene, representation shouldn’t be left to an algorithm. Casting, particularly, is a decision that deserves conscious human judgement.

The big question, then, is not whether AI can do more. It can. The question is whether we will know when to step in.

AI in Film

We’re still working out the language around new ways of doing things. But it’s perhaps a little reductive to talk about ‘AI film’ and more constructive to talk about AI in film?

Maybe it’s not about ‘keeping the human in the loop’ – but how we best use AI in our process to ensure quality of output?

So let’s keep asking, where does the human do their thing best? Where does AI do its best? Let’s bring it in at the right moments, for the right reasons. Let’s be mindful of surrendering creative choices and be conscious of what will be sacrificed if we do.

In our world, credibility and authenticity matter enormously. So let’s always ask where the humanity in a project will come from. And, if the use of AI will contribute or counter it.

AI tools offer mind-blowing possibilities – both creatively and in reducing bottom line. But going forward, the best creative, the best ideas, will be more important than ever. Without an idea, a point of view, some humanity and a burning desire to connect – we risk contributing more flotsam and jetsam to the sea of slop mediocrity.

Originality – the thing that surprises us, challenges us and moves audiences – still begins with people. Our best AI-enabled work to date has not been because of the tech – it’s been because of those using it.