grantmaking.ai Launch Round
Who we are
We create films, audio plays, and videos about risks from transformative AI, often made using cutting-edge AI-video tools. We pursue two tracks: (i) short films and other videos for social media, and (ii) development of flagship feature films and audio plays.
We’re currently a two-person team: Keiran Harris (Creative Director, former Podcast Producer and Head of Audio at 80,000 Hours), and Chloe Harris (Operations Lead, former Head of Systems at Effective Ventures).
What we’ve done so far
We’ve written the full screenplay for our first feature-length film, “Last Year, Baby!”, choosing it as the best option out of a few ideas and possible outlines we developed. The movie follows the days leading up to a Senate vote to determine the future of superintelligent AI, as two rival CEOs and a White House Chief of Staff are each pretty sure they should be the most powerful person who ever lived.
The script is built around three goals:
-
Be a legitimately entertaining movie that a broad audience would actually want to watch.
-
Introduce the risks inherent in the race to Artificial Superintelligence, both through the dialogue and what happens, such that a viewer with no prior exposure could feel the problem viscerally by the end of the movie.
-
Minimise the risks of the failure modes that would make it counterproductive: partisanship, preachiness, or the sense that what happens is purely sci-fi fiction.
While there have been lots of movies made about AI down the years, and lots more on the way, we’re not aware of another screenplay attempting this for the race to ASI.
We also created four experimental “mini movies” (using Veo 3 / Midjourney) including The First Intelligence Explosion, which got 98K views and dozens of positive comments on Twitter / X.
We were pleased with the quality of these given the tools at the time. But after asking “if Alien Cub is a wild success in a few years, how did we get there?”, we decided the expected impact of a feature-length script was higher. And if we were ever going to write a movie, it made sense to do that while the technology was clearly not ready, rather than waiting for the key breakthroughs first.
What we’ll do next
1. Produce and release ‘Last Year, Baby!’ as a podcast series
Now that the screenplay exists, our next step is to turn Last Year, Baby! into a polished audio play, released publicly as a multi-part podcast series. We think there are a few nice things about this plan:
- It doesn’t matter how good your screenplay is unless people actually read it. Asking folks who’ve never read a screenplay to work through your 100+ page movie is a somewhat big and weird ask, but listening to a podcast is a pretty easy ask, and for most people it’ll probably be a more fun introduction to the work. Getting your work in front of the right people is key to actually getting a movie made via traditional means, and a public audio series increases those chances vs. a screenplay.
- Though we’d have a strong preference to get the movie made in real life with real actors, that might not be possible. But AI video tools might progress to the point where it’ll be possible to create a good version on our own (to be clear, the current tech isn’t anywhere near this today, we just think it’s plausible the tech will get there, especially for animated / anime films). A professional audio play could serve as the near-finished audio track for an AI-led movie, thus speeding up film production massively in the event the tools progress sufficiently.
- I (Keiran) produced The 80,000 Hours Podcast between 2017-2024, so can do everything associated with releasing it as a podcast at a high level.
- Scripted audio series → TV shows or movies is now an established production pipeline, so we could follow in the footsteps of success stories such as Gimlet Media’s Homecoming (which became a critically acclaimed Amazon TV series).
2. The Last Show
I’ll also produce “The Last Show”: a recurring fictional late night show monologue (like David Letterman, or Conan O’Brien) dedicated to the latest AI news. This seems meaningfully different vs. existing newsletters and podcasts that cover the latest in AI, and could appeal to a different audience. Once I’ve made one successful episode, it’ll be relatively straightforward to reproduce them regularly with current news stories / monologue jokes.
3. Create polished, standalone, short films / videos
Depending on funding / time, we’ll also aim to create regular standalone videos that cover especially relevant topics or concepts, experimenting in different styles as we’ve already done with existing videos.
Dr. Strangelove showcased the insanity of nuclear war to the point it became a reference point in the 1964 US election, and we started “Last Year, Baby!” hoping to write “Dr. Strangelove, but for AI”. We ultimately decided to write a different type of dark comedy, but the goal remains of creating the movie about the race to superintelligent AI (as well as the scripted audio series, but that just doesn’t have the same ring to it).
The Day After, another inspiration, is seen as a big deal for convincing Reagan of the dangers of nuclear war, and we hope it’s possible to influence politicians, from those working in relevant policy positions to world leaders. But alerting millions of regular people who’d otherwise never hear the core case for worrying about AI x-risk would also be a big win in expectation in terms of passing sensible legislation.
Both films worked because they were first and foremost good movies. Our flagship film, Last Year, Baby!, follows through on our bet that something similar is possible for AI, with the podcast series designed to test that idea.
We think of the feature film path as having asymmetric upside, but building a thriving YouTube channel of >1M combined views is a more conventional path to success that we’ll pursue simultaneously.
The minimum amount ($15k) funds one month of running costs, during which we’ll produce a pilot for the Last Year, Baby! podcast series (combining human voice performance with AI alterations).
The max for the round ($50k) funds ~3 months of operational costs, during which we’ll produce the full podcast series and start releasing episodes, with human performers for principal roles if the hybrid model disappoints. We’ll also produce and release the first episode of The Last Show, as well as at least one standalone video.
The ideal amount ($190K) secures funding for 12 months. That covers staff, operational (e.g. company insurance, payroll administration), and project expenses (e.g. software and subscriptions, marketing, voice talent), during which we’ll produce and release the entire podcast series, regular episodes of The Last Show, monthly standalone videos, and push forward on developing the feature film.