Jason Ferrell

AI product builder, fCAIO, AI film director

[Roger Avary @AVARY]

With all the doomsayer talk of the death of Hollywood, let’s put things into perspective. This has been a great year. My friends in the exhibition business are downright upbeat, something they haven’t been for years. Audiences are going to the movies. They say 2027 and 2028 will be even better. We’ve just seen two indie films from new directors do huge business, thanks to skillful & deft Hollywood distribution. Disclosure Day has done respectable business and like the film or not, it had people talking. And even if you didn’t care for it’s boomer perspective, it’s coming on the heels of Project Hail Mary, which was Hollywood doing what Hollywood does best, with traditional positive storytelling, also with aliens, and a pro humanity, pro morality, visually fantastic character study which explores our relationship with God. We have a new Paramount rising from the brink with the promise of 30 new films a year. There’s new and exciting storytelling tools available to all, heretofore unseen in history, which will certainly advance the medium. The year 2026 has been awesome so far. I can’t wait to see what’s next to come. Hollywood’s absolutely not dead.

[Jason Ferrell @jtferrell512]

@ChrisGwinnLA 4 takes on average for most shots, but it depends on the complexity of the shot, complex action could be 10 or so … but that’s just the video side, anchor and end frames often need multiple iterations before you even get to the takes.
64 takes per shot is wild

[Jason Ferrell @jtferrell512]

@rachel_l_woods Even more frightening is how behind the times the AI consulting firms ostensibly shepherding those CEOs to modernity are

[Jason Ferrell @jtferrell512]

@JoshDaws Nice! I made a similar 3d blocking tool but with gpt-5.2 at the time it took a hell of a lot more than one prompt!

[Jason Ferrell @jtferrell512]

I set a few /goal threads in codex before the Memorial Day weekend - they’re still running 100+ hrs later and making real progress. AGI is pretty much here.

[Jason Ferrell @jtferrell512]

@jennykrakovsky Nice work! You might be able to get the flicker and sound to converge on something good in DaVinci Resolve - just part of the limitations and opportunities of the art form for now

[Big Brain AI @realBigBrainAI]

Peter Steinberger, creator of OpenClaw, on why AI agents still produce "slop" without human taste in the loop:

"You can create code and run all night and then you have like the ultimate slop because what those agents don't really do yet is have taste."

Peter is direct: raw capability without direction still produces mediocre output.

"They are spiky smart and they're really good at things, but if you don't navigate them well, if you don't have a vision of what you're going to build, it's still going to be slop. If you don't ask the right questions, it's still going to be slop."

Great AI-assisted work is defined by the human guiding it.

@steipete describes his own creative process when starting a new project:

"When I start a project, I have like this very rough idea what it could be. And as I play with it and feel it, my vision gets more clear. I try out things, some things don't work, and I evolve my idea into what it will become."

Most people skip this part entirely, front-loading everything into a single prompt and wondering why the result feels hollow.

"My next prompt depends on what I see and feel and think about the current state of the project."

Each step informs the next. The work itself is the feedback loop.

"But if you try to put everything into a spec up front, you miss this kind of human-machine loop. And then I don't know how something good can come out without having feelings in the loop — almost like taste."

The agentic trap is what happens when you remove yourself from the process too early.

[Andrej Karpathy @karpathy]

Something I've been thinking about - I am bullish on people (empowered by AI) increasing the visibility, legibility and accountability of their governments.

Historically, it is the governments that act to make society legible (e.g. "Seeing like a state" is the common reference), but with AI, society can dramatically improve its ability to do this in reverse. Government accountability has not been constrained by access (the various branches of government publish an enormous amount of data), it has been constrained by intelligence - the ability to process a lot of raw data, combine it with domain expertise and derive insights. As an example, the 4000-page omnibus bill is "transparent" in principle and in a legal sense, but certainly not in a practical sense for most people. There's a lot more like it: laws, spending bills, federal budgets, freedom of information act responses, lobbying disclosures... Only a few highly trained professionals (investigative journalists) could historically process this information. This bottleneck might dissolve - not only are the professionals further empowered, but a lot more people can participate.

Some examples to be precise: Detailed accounting of spending and budgets, diff tracking of legislation, individual voting trends w.r.t. stated positions or speeches, lobbying and influence (e.g. graph of lobbyist -> firm -> client -> legislator -> committee -> vote -> regulation), procurement and contracting, regulatory capture warning lights, judicial and legal patterns, campaign finance... Local governments might be even more interesting because the governed population is smaller so there is less national coverage: city council meetings, decisions around zoning, policing, schools, utilities...

Certainly, the same tools can easily cut the other way and it's worth being very mindful of that, but I lean optimistic overall that added participation, transparency and accountability will improve democratic, free societies.

(the quoted tweet is half-ish related, but inspired me to post some recent thoughts)

[Andrej Karpathy @karpathy]

- Drafted a blog post
- Used an LLM to meticulously improve the argument over 4 hours.
- Wow, feeling great, it’s so convincing!
- Fun idea let’s ask it to argue the opposite.
- LLM demolishes the entire argument and convinces me that the opposite is in fact true.
- lol

The LLMs may elicit an opinion when asked but are extremely competent in arguing almost any direction. This is actually super useful as a tool for forming your own opinions, just make sure to ask different directions and be careful with the sycophancy.

[Jason Ferrell @jtferrell512]

@arlogilbert If it took a year to build, they should have already encountered this problem in a big way at least 3 times and solved for it one way or another.

[Jason Ferrell @jtferrell512]

@minhsmind I built it

[Jason Ferrell @jtferrell512]

@Diesol Very cool! But as a guitar player it’s a little cringy watching the fingers do something different from the music - same is true for live action movies where the actor can’t really play the instrument

[Jason Ferrell @jtferrell512]

@aedison I designed my AI director assistant to follow the 180 rule

https://t.co/axYCrbGvPp

[Jason Ferrell @jtferrell512]

Here’s Act 1 of a feature film made entirely with AI.

The second half was made with Kling 3.0, the first half with the vastly inferior Gemini Veo 3.1.

Everything about Kling was better and easier to work with, especially since Google’s safety filters have gotten significantly more strict and unpredictable.

While the Veo 3.1 half is barely watchable, maybe useful for previz, the Kling half makes me feel like we have entered the era where a low budget horror film people would pay to see in a theater can be made by one person with AI.

https://t.co/oKHJaOWD4l

[Jason Ferrell @jtferrell512]

@GlennHasABeard I adapted the first act from a classic screenplay with a unique open copyright situation, Night of the Living Dead, to AI … horror isn’t my favorite genre, but its a nice challenge to build tension to a suspenseful action sequence with AI

https://t.co/axYCrbGvPp

[Jason Ferrell @jtferrell512]

Here’s Act 1 of a feature film made entirely with AI.

The second half was made with Kling 3.0, the first half with the vastly inferior Gemini Veo 3.1.

Everything about Kling was better and easier to work with, especially since Google’s safety filters have gotten significantly more strict and unpredictable.

While the Veo 3.1 half is barely watchable, maybe useful for previz, the Kling half makes me feel like we have entered the era where a low budget horror film people would pay to see in a theater can be made by one person with AI.

https://t.co/oKHJaOWD4l

[Jason Ferrell @jtferrell512]

Great analysis!

I’d love it if you did a part 2 on the second half of my Night of the Living Dead adaptation where I used a better model (Kling 3.0) and incorporated your feedback from the first half

It’s mostly an action sequence, which presented different challenges from the slow-burn dialogue sequence in the first half

https://t.co/NoMB4uOW58

[Jason Ferrell @jtferrell512]

Here’s Act 1 of a feature film made entirely with AI.

The second half was made with Kling 3.0, the first half with the vastly inferior Gemini Veo 3.1.

Everything about Kling was better and easier to work with, especially since Google’s safety filters have gotten significantly more strict and unpredictable.

While the Veo 3.1 half is barely watchable, maybe useful for previz, the Kling half makes me feel like we have entered the era where a low budget horror film people would pay to see in a theater can be made by one person with AI.

https://t.co/oKHJaOWD4l

[Jason Ferrell @jtferrell512]

@Artedeingenio Curious your thoughts on my AI adaptation of Night of the Living Dead Act 1 - I upgraded to Kling 3.0 for the second half

The first half is intentionally slow paced with long shots to build tension for the second half

https://t.co/axYCrbGvPp

[Jason Ferrell @jtferrell512]

Here’s Act 1 of a feature film made entirely with AI.

The second half was made with Kling 3.0, the first half with the vastly inferior Gemini Veo 3.1.

Everything about Kling was better and easier to work with, especially since Google’s safety filters have gotten significantly more strict and unpredictable.

While the Veo 3.1 half is barely watchable, maybe useful for previz, the Kling half makes me feel like we have entered the era where a low budget horror film people would pay to see in a theater can be made by one person with AI.

https://t.co/oKHJaOWD4l

[Jason Ferrell @jtferrell512]

Thank you. I think you’d be pleasantly surprised the amount of decision making and improvisation involved in producing something like this. Every shot started with an intent, the AI gave me what it gave me, then I had to react to that, perhaps the same way a director might react to an actor doing something unexpected.

[Jason Ferrell @jtferrell512]

This screenplay has a unique copyright situation that makes it great as a case study - sorry it had to be horror.

The classic script may be hard to follow, sure … (they are brother and sister, not married)

But hopefully this shows how one can adapt a screenplay to a movie as a low budget AI production and how that compares to low budget production from 1968. Personally, I think the AI acting is better than the 1968 version.

[Jason Ferrell @jtferrell512]

@TaraLynnSays Have you seen the original 1968 version? How does it compare to you?

[Jason Ferrell @jtferrell512]

@ZackMorrison18 Full act 1 version with updates: https://t.co/axYCrbGvPp

[Jason Ferrell @jtferrell512]

Here’s Act 1 of a feature film made entirely with AI.

The second half was made with Kling 3.0, the first half with the vastly inferior Gemini Veo 3.1.

Everything about Kling was better and easier to work with, especially since Google’s safety filters have gotten significantly more strict and unpredictable.

While the Veo 3.1 half is barely watchable, maybe useful for previz, the Kling half makes me feel like we have entered the era where a low budget horror film people would pay to see in a theater can be made by one person with AI.

https://t.co/oKHJaOWD4l

[Jason Ferrell @jtferrell512]

Prof. Zack, I’m teaching myself filmmaking with the classic George Romero screenplay Night of the Living Dead. I’m experimenting with long shots to build tension at horror pacing.

Curious if you might be so kind as to offer any notes on the effort so far. Veo 3.1 already feels ancient and quite limiting, but excited about new models on the horizon.