a collectively-authored anthology of futures during and after the intelligence explosion, by and for the winter 2026 AI safety community.
Drastically different futures are likely near. We think safety researchers need to be imagining those futures – endpoints and waystations, across the full spectrum of how things might go – with clarity, specificity, and creativity.
The landscape of futures storytelling remains surprisingly narrow relative to the sheer uncertainty and diversity of possible outcomes that lie ahead. This remains true within the alignment community. Despite a robust tradition of writing and fiction as a means of community and idea-building, most narratives come from just a few authors.
We'd like to guard against the atrophy of imagination. As the next generation of AI safety researchers and leaders, we want to cultivate our own ability to think concretely about futures, and to share the perspectives we develop.
Proxima is a project by and for the winter 2026 safety community: MATS, Astra, and AFP fellows, mentors, staff, and adjacent researchers. It will be a collectively-authored anthology of prose, poetry, and visual art about plausible, well-considered futures during and after the intelligence explosion, published to coincide with the end of our programs and containing the futures our cohort imagines.
We see three reasons to do this project:
The discourse needs more imagination. There is too little diversity and plurality in how possible futures are imagined in the broader AI conversation. Most visions cluster around a few familiar scenarios. We want to expand the range, and in doing so steer towards the futures that we want, as Neal Stephenson's Snow Crash (1992) did for the metaverse or William Gibson's Neuromancer (1984) did for cyberspace.
Our perspectives are worth sharing. Our cohorts are remarkably diverse in background, identity, and viewpoints. Simultaneously, we together represent a large chunk of the future of AI safety researchers and leaders. What we think about the future is valuable to both other fellows and members of the wider community; we have something to say.
Training the imagination muscle makes us better researchers. The creative practice pushes us to lay out details, contradictions, and areas of promise and problems that don't surface easily through our technical or policy projects research alone. We see Proxima as a collective exercise in sharpening instincts that we can bring back to our main research agendas.
We hope that by increasing the pluralism and specificity of the futures we envision, we empower ourselves and our community to act on those visions – and to help the future go more well.
We're seeking prose, poetry, and visual artwork about plausible futures in and after the intelligence explosion.
This call is primarily for current MATS, Astra, and AFP fellows, but the broader fellowship community is also welcome to submit: mentors, alums from previous cohorts, staff and support members, and researchers at organizations in Constellation and Lighthaven.
We want work that takes its future seriously and makes it concrete, specific, and alive. These futures need not be obviously good or obviously bad. We're interested in worlds that are odd, mundane in unexpected ways, unsettling, tender, strange, funny. We're interested in a range of scopes: from the global to the granular – a single human (or post-human! or non-human), a discipline, a website, a cultural practice, a technology, a meme.
Beyond that, our call is very open. Take whatever direction feels right to you.
What we're not looking for: AI 2027-style strategy scenarios, policy briefs, or fantasy fiction set in worlds that seem obviously impossible.
Short fiction (up to 5,000 words)
Poetry (up to 3 poems per submission)
Visual artwork (illustration, photography, mixed media: 300 DPI minimum, PNG or TIFF for print)
Wildcard: We welcome submissions that don't fit the categories above: video games, storyboards, software, dance, digital/multimedia art, static websites, interactive pieces, or anything else you can imagine. We'll work with you to publish it in a format that's true to your vision.
Deadline: March 20, 2026
Submit to: submission form
Note: Proxima will also be organizing ~3 coworking sessions, alternating between Constellation and Lighthaven, to co-worldbuild with other fellows and support your submissions. Stay tuned.
Err toward submitting rather than not. We want to hear your voice.
We appreciate and encourage high-quality work, but this is a community project – what matters most is that you engaged with these futures and put something on the page or the canvas. Length doesn't matter: we welcome very short pieces as well as longer ones. We're particularly enthusiastic about submissions from people with no prior experience in fiction, poetry, or visual art.
We know fellows are busy with their own research. Think of this as a complement to that work, not a distraction from it – we've found that imagining futures concretely and creatively sharpens the thinking we bring back to our main projects. We (Jasmine and Parv) are carving time out of our own fellowships to organize this because we think it's worth it, and we'd be grateful for whatever you're able to contribute.
We will select samples of written pieces and visual artwork to assemble a printed anthology. ~200 high-quality printed copies of this anthology will be distributed to the AI safety community at the end-of-program MATS exhibition on March 28, 2026.
In addition, all submissions will also be published online, on proxima.ink. We aim for this to be a beautiful website and to show your work generously.
For tone, scope, and ambition, we admire:
Richard Ngo's The Gentle Romance; nostalgebraist's Almost Nowhere; Scott Alexander's Unsong
Ted Chiang's The Lifecycle of Software Objects, Exhalations
Exhibitions at the Misalignment Museum, SF
Iain M. Banks's Culture novels
FLI's worldbuilding contest
Better Images of AI
Art by Spencer Chang
The Walker Art Center's Designs for Different Futures exhibition
Stewart Brand's Whole Earth Catalog