Film festival programming is one of the most human activities in the screen industry. A group of people watch hundreds or thousands of films and, based on their taste, experience, and instinct, select a program they believe represents the best and most interesting work available. It’s subjective, it’s personal, and it’s the reason different festivals have different personalities.

So when festivals start using AI tools in the programming process, it raises important questions. What role should technology play in deciding which films audiences see? And what gets lost when algorithmic analysis enters a fundamentally human process?

What’s Actually Happening

Let me be clear about what I’m talking about. No major film festival that I’m aware of is using AI to make final programming decisions. What several festivals are doing is using AI-assisted tools for the earlier stages of the submission and screening process.

The most common application is submission triage. Major festivals receive thousands of submissions, far more than their programming teams can physically watch in full. AI tools can analyse metadata, synopses, and submission materials to help prioritise which films get watched first by human programmers. This is essentially an efficiency tool, reducing the administrative burden of managing large submission volumes.

Some festivals are also using AI tools to analyse audience data from previous editions to inform programming decisions. If data shows that certain types of films consistently perform well with the festival’s audience, that information can be used alongside the programming team’s creative judgment.

A handful of festivals have experimented with AI analysis of the actual film content, using computer vision and audio analysis to assess technical quality, pacing, and other measurable characteristics. This is the most controversial application and, from what I understand, the results have been mixed at best.

The Australian Context

Australian festivals, with smaller submission volumes than Sundance or Toronto, have been slower to adopt AI tools. But even at the mid-tier level, festival programmers are managing hundreds of submissions with limited staff, and some are exploring whether technology can help.

MIFF, SFF, and AFF haven’t publicly discussed AI tool adoption in their programming processes, and I suspect their approach is appropriately cautious. The strength of these festivals is their curatorial vision, which is inherently human.

Smaller festivals, which often rely on volunteer programmers, might benefit most from AI-assisted triage tools that help manage the submission workload without replacing the human decision-making that defines the program.

The Concerns

The primary concern with AI in festival programming is that it could homogenise programming decisions. If AI tools are trained on data about what’s been successful in the past, they’ll tend to favour work that resembles past successes. That’s precisely the opposite of what good festival programming should do. The best festivals discover something new, something unexpected, something that challenges the audience’s assumptions.

The most interesting films at any festival are often the ones that wouldn’t score well on any algorithmic assessment. They’re the formally unconventional, the culturally specific, the genre-defying work that a human programmer recognises as significant through their expertise and intuition.

There’s also a fairness question. If AI triage tools deprioritise submissions based on metadata rather than content, films from less established filmmakers, less familiar countries, or less conventional genres might never reach the human programmers who could champion them.

Where AI Could Genuinely Help

I’m not opposed to AI tools in festival operations. There are applications where they could be genuinely useful without threatening the curatorial integrity of the programming.

Logistics management: scheduling screenings, managing venue availability, and optimising session times for audience flow are complex logistical tasks where AI could improve efficiency.

Audience analysis: understanding who attends which types of screenings and using that data to inform marketing and scheduling decisions is a legitimate application that doesn’t affect creative programming.

Accessibility: AI-powered transcription and subtitle generation could make festival programs more accessible to audiences with hearing impairments or language differences.

Administration: managing submission databases, tracking correspondence with filmmakers, and processing applications are all administrative tasks where AI could reduce workload.

Some festivals have been working with an AI consultancy to develop approaches that use AI for operational efficiency while keeping creative decisions firmly in human hands. That seems like the right framework.

The Bottom Line

Film festival programming should remain a human activity. The value of a festival is in the taste, knowledge, and courage of its programming team, qualities that AI cannot replicate. But AI tools can make the operational side of running a festival more efficient, freeing up human time and attention for the creative work that matters.

The festivals that get this balance right will be stronger for it. The ones that let AI creep into creative decision-making risk losing the very thing that makes them worth attending.