AI Films and the Audience Gap

AI tools make it easier to turn a screenplay into a moving draft. Finding an audience is a separate problem. The useful question is how far easier production has translated into films people choose to watch, discuss and return to.

This matters to the way I am making 5RPS. I am continuing a full AI-assisted draft so I can see and hear the screenplay from beginning to end, then judge what drawing, performance, sound and funding it still needs. TEGAKI grew out of an earlier experience with AI backgrounds and the reactions around them. The production question and the audience question have stayed with me. Director commentary

The records below were checked on October 17, 2026. This article was revised on October 18. Willingness changes with the use described, and some generated films have reached paying audiences. The records examined here cannot establish sustained demand. “Audience gap” names that unresolved question, not a calculated difference between two populations. Evidence comes from surveys in Great Britain and the United States, ordinary cinema records, and a small self-selected poll. These sources answer different questions and are not combined into one audience estimate. AI-film festivals, awards and creator-community engagement are excluded from the reception evidence because their selected audiences cannot establish general acceptance.

Different meanings of making it

General-public surveys show both resistance and willingness

YouGov asked 5,382 adults in Great Britain whether knowing one character in a live-action film was AI-generated would change their likelihood of watching: 60% said less likely, 21% said no difference, 5% more likely, 8% were unsure and 6% said they do not watch movies. This is about one disclosed AI actor, not a fully AI-made film. YouGov GB AI actor poll

A separate Angus Reid USA survey of 2,007 adults asked about willingness to watch TV or movies made entirely by AI, and then partly by AI and partly by people. The entirely AI-made version drew 11% willing without reservations, 28% willing with reservations, 50% not willing and 11% unsure. For partly AI, partly human work, the figures were 15%, 37%, 39% and 10%. The partly AI row totals 101% because published rounded percentages are retained. Angus Reid AI entertainment poll

The task also changes the answer. YouGov's separate October 2–5, 2025 survey of 1,106 U.S. adult citizens reported 61% accepting AI visual effects, against 18% accepting an entire film made without human actors or writers. Its reported margin of error was ±4 percentage points. This is acceptability, not willingness to watch, and is not a trend comparison with the later Angus Reid question. YouGov poll tables

A small poll on a software-engineering page

A poll on an older Remote Startup Senpai page, built for a software-engineering audience, asked for reactions to a Seedance animation reel made from a manually drawn layout. The exact question was ‘thoughts on AI slop?’

This is a self-selected response to one reel and a negatively framed question, not a general-public estimate. It cannot isolate the effect of the AI label or predict how people respond to a completed film.

Remote Startup Senpai: colored-in roughs

Colored-in roughs / grid 1

Colored-in roughs / grid 2

Cinema bookings, reception and money

Bookings, admissions and reception answer different questions. Each case keeps its finding visible and its dated exact record available below. Small platform aggregates are source records, not audience statistics.

These records show uneven reception. Post Truth records 1,983 admissions over nine weeks; the live-action hybrid Run to the West has a larger reported audience of about 28K. Admissions measure attendance, not approval. Small platform records for generAIdoscope, 怪奇！死肉の男 and Love You contain favorable and unfavorable responses. No title supplies the share of all viewers who liked it. A ticket price without admissions supplies no gross, and gross without costs supplies no profit.

What appeared under eight theatrical trailers

On October 17, 2026, I recorded visible counters and up to the first ten non-uploader, top-level comments shown by default under eight ordinary theatrical trailers: five AI/hybrid productions and three artist-animated films. No replies were expanded. I coded each displayed comment by its expressed response: positive for praise, anticipation or support, negative for criticism or rejection, mixed for both, and other for questions, predictions, jokes or unclear/off-topic remarks. This convenience sample is not representative of the general public and does not match countries, genres, ages or budgets. Trailer UI records Trailer comment coding

The useful part of this collection is the coexistence of different arguments. Love You attracted encouragement for a Kannada filmmaking experiment as well as criticism of the images. ChaO attracted criticism of voices and design as well as praise for voice performances. These are examples of expressed reactions, not a test of which production method audiences prefer. A comment praising an attempt does not necessarily endorse the finished feature. Silence does not establish indifference or approval. Trailer comments are separate from feature reviews and admissions. ChaO official site Last Blossom official site Trailer UI records Trailer comment coding

Can an external tool verify the audience?

A large view counter and few comments can coexist in either production group. The public record does not resolve how many people finished, returned or paid. YouTube's invalid-traffic assessment is separate from these public measurements. Invalid traffic help

How to read this evidence

This is a bounded descriptive inquiry, not a systematic review or a causal study. It connects accessible surveys, cinema records and selected public reactions available in the October 17 search. The cases were not selected randomly or declared before searching. Their inclusion documents access to a source, not their representativeness. Missing evidence remains missing.

What the record can answer

The surveys show substantial resistance alongside a sizable willing minority. They do not measure completed viewing, purchases or repeat viewing. This search did not establish sustained mainstream demand for AI-made features, and does not establish its absence elsewhere. The cinema records and self-selected poll add context, but cannot fill those gaps or support a conclusion about Japan or global audiences.

Why TEGAKI, and why keep building 5RPS

I made TEGAKI after working with artists on an earlier 5RPS trailer and encountering the response to AI-assisted backgrounds. Consent, creative work and friendship became the short's subject. The longer account is in the director commentary. Director commentary

5RPS is still being built as a complete AI-assisted draft so the screenplay can be seen, heard and judged from beginning to end. The result can show what still needs better animation, performance, sound or funding. The aim is to finish it and put it out there. A finished draft might eventually help secure support for a version made with professional artists. That support and the final production method remain undecided.

Research note: sources and earlier observations checked October 17, 2026. AI assisted searching and organization, followed by source checking and editing. Ratings, creator claims, admissions and private-page observations are attributed separately. Other productions' private analytics and finances were not inspected. The cover is a newly generated Liz illustration, not a completed 5RPS frame.