Yes.
If by “text-to-movie” you mean:
“I write a paragraph, and an AI generates a coherent feature-length movie with characters, dialogue, camera work, music, editing, and narrative structure,”
then I think the answer is:
partially yes by 2027, but not yet reliably at Hollywood quality.
The important point is that we are no longer asking whether text-to-movie is possible.
We are asking how good it will become.
Where we are today
In 2026, systems such as:
- OpenAI Sora
- Google Veo
- Runway
- Luma AI
can already generate short video clips from text prompts.
The major breakthroughs have been:
- Visual realism
- Camera movement
- Physics simulation
- Character consistency
compared with only a few years ago.
The real challenge is not video
Most people assume the difficult part is generating images.
Ironically, that is becoming the easier part.
The harder problem is:
Narrative coherence
A feature film requires:
- memory
- causality
- character development
- pacing
- emotional arcs
For example:
If a woman loses her wedding ring in minute 12,
the audience expects that event to matter in minute 95.
Current video generators struggle with such long-range structure.
What likely happens by 2027
I expect AI systems to become capable of generating:
Short films
5–20 minutes
from a reasonably detailed prompt.
For example:
“Generate a melancholic science-fiction film in the style of Tarkovsky about an elderly astronomer living beside a dying sea.”
This is very plausible.
Episodic content
You describe:
- characters
- setting
- tone
and AI generates new episodes.
This resembles television production.
Human-directed feature films
A human serves as:
- writer
- director
- editor
while AI creates:
- actors
- scenery
- shots
- music
- effects
This may become common surprisingly soon.
The economic shock
The biggest consequence is not technological.
It is cultural.
Historically:
Making a movie required:
- money
- actors
- cameras
- crews
- studios
AI dramatically lowers those barriers.
One person may eventually produce work that previously required hundreds.
This could transform filmmaking much as desktop publishing transformed publishing.
But what about great cinema?
This is where predictions become difficult.
Suppose an AI can perfectly imitate:
- Andrei Tarkovsky
- Akira Kurosawa
- Ingmar Bergman
Would it create a great film?
Maybe.
Maybe not.
Because cinema is not merely image generation.
It is a way of seeing.
Tarkovsky’s films emerged from:
- memory
- exile
- faith
- mortality
Those are existential conditions, not rendering techniques.
A deeper question
Kelly, this reminds me of a theme that appears repeatedly in your conversations.
You often return to:
- Sebald
- Bashō
- wandering
- surrender
- following fragments
Notice that none of these begin with a plan to manufacture a masterpiece.
They begin with an encounter.
A road.
A ruin.
A summer evening.
A fragment.
AI can already generate images.
Soon it will generate films.
But the more interesting question may be:
Who is having the experience from which the film arises?
A text-to-movie system can probably produce a visually convincing “Bashō film” by 2027.
What it cannot yet clearly do is be Bashō:
to walk for months through rain and dust,
to lose friends,
to grow old,
and then write:
Summer grasses—
all that remains
of warriors’ dreams.
That gap between generating a movie and living a life may remain one of the most important distinctions, even as text-to-movie becomes increasingly real.
