Taking jaggedness seriously ⊗ All futurists are archivists

No.382 — Collapse ⊗ Designing responsible futures ⊗ The medical case for self-driving cars ⊗ Mushrooms in recycled egg carton towers ⊗ 100 notable small press books

Taking jaggedness seriously ⊗ All futurists are archivists
Fluid dynamics (see jaggedness piece).

Taking jaggedness seriously

Strangely enough, I’ve never mentioned the concept of AI “jaggedness,” even though it’s been around for a little while. This written version of her talk by Helen Toner is an excellent occasion to do so. Imagine every task humans do placed in a circle, easy in the center, and progressively harder as you get further from the center. Now draw a circle somewhere in there, a frontier or level of difficulty. Jaggedness represents the observation that for LLMs some tasks that seem very hard to us are easily done, while some super easy ones stump them. Imagine a wavy unequal shape that intersects your circle constantly, that’s the weird contour of LLM capabilities. Here’s the orignal paper on the concept and here’s the slide from Toner’s deck representing it.

She argues that AI progress will remain jagged, systems will keep getting much better at some tasks while staying weak at others, and that unevenness will matter for deployment and disruption. Toner explains that jaggedness arises from context‑specific application challenges, differences between cognitive and physical tasks, adversarial or safety constraints in certain industries, and the social/relational skills humans bring to their work/tasks. The main point of her talk, to me anyway, is that rather than fixating on distant endpoints like AGI, she encourages us to focus on the varied pace and intensity at which different AI capabilities develop and are applied.

I think we could be heading for a world where we don’t necessarily skip to the end and AI is automating everything; instead, humans are still in many positions of power, positions of influence, but they are increasingly collaborating with AI or getting advice from AI.

Reading the above, I started thinking of “humans in the chair.” We usually talk about keeping the “human in the loop,” which looks a lot like a human failsafe. LLMs do the work but there’s a spot where humans validate. What if instead it’s still humans everywhere, but with superintelligent calculators? Obviously, that’s nothing new as a potential outcome, I just like the broader “control chair” vs the monkey in the loop.

Timelines to AGI crazy: For a long time, I have not found a very helpful to talk about timelines to AGI, because I think people use just such different metrics and definitions for what they mean by that. I tend to think more in terms of how crazy are things going to get, how soon? And you can use different metrics for crazy, and I find that maybe a more productive and more interesting way to think about things. […]

Focusing less on, “Well, trends are going this way, so why don’t we just skip to the end and assume that we’re there?” And instead being more curious about the order in which we get different capabilities. Being more curious about the speed and the time differences between when those different capabilities arrive. And being more curious about kind of the intensity at which we can exploit different capabilities. So some things are really working well and are easy to just dial up 100,000x, and other things are still a grind and getting even 1x is a challenge. And that’s going to give you jagged deployment, jagged implications of AI as well. […]

I tend to think that there’s instead going to be a big premium for the companies and organizations that can figure out how to adopt AI. And that’s going to mean changing processes, doing things afresh. It’s going to advantage new entrants who don’t have a lot of institutional cruft. And that’s going to disrupt who is in power and therefore who is steering and who is making decisions.

All futurists are archivists, all archivists are futurists

Loved this piece by Amahra Spence. Thinking about the overlap between futurism and archiving, she realised that they are actually the same discipline, each concerned with care, maintenance, and survival. Spence is focused on the Black experience but it’s a valid case for any futures work, and actually has deeper meaning here.

In Black contexts, systemic neglect has been structured into the environment itself—schools closed, neighbourhoods demolished, records disappeared. This makes archiving an infrastructural necessity rather than a creative pursuit. Black life has required temporal dexterity: the ability to look backward and forward simultaneously, carrying suppressed histories while building new worlds despite rupture. The Sankofa principle makes this explicit—retrieval of what was taken becomes a precondition for moving forward, a political stance that refuses the idea of progress as forward motion detached from the past.

The archives that matter operate through a different logic than institutional heritage spaces. They require touch, breath, tactility, conversation, and risk rather than sterile storage behind glass. This tactile, embodied practice counters structural abandonment. Care interrupts erasure; it makes continuity possible. The work prioritises maintenance over extraction, tactility over sterility, collective governance over gatekeeper authority. Futurism divorced from memory work becomes amnesia, risking the recreation of harm under the guise of innovation. Both practices refuse erasure and construct infrastructures through which Black life can thrive.

Dr. Etienne Joseph often reminds us that archival work is never neutral. It begins long before a record is catalogued or a photograph is digitised. It begins in the care that someone extends to a memory others might dismiss. Care is what interrupts erasure. Care is the counterforce. Care is what makes continuity possible. […]

Futurism is not an escape from the past; it is the stewardship of the past into new conditions. It’s how we draw from our histories to demand and bring into being futures in which our lives matter. […]

• maintenance over extraction
• tactility over sterility
• intergenerational memory-work over erasure
• stewardship over spectacle
collective governance over gatekeeper authority
• community benefit over institutional prestige
• long-term stability over short-term visibility

Collapse

Rosie Spinks explores what she calls becoming “collapse aware,” which means acknowledging that our current way of life may be nearing its end and that infinite progress is not guaranteed. She emphasizes the emotional aspect of living through collapse, suggesting that it necessitates a shift in our values and priorities, focusing more on sustenance rather than status. She presents collapse awareness not as a grim outlook but as a way to energise and prepare for an uncertain future while fostering connection and care. (It’s from last year and she recently wrote a follow up.)

Also on collapse, Luke Kemp was interviewed on the Farsight podcast about his book Goliath’s Curse. It’s much bleaker and doomy than Spinks’ piece, but also goes pretty deep on the history of civilisational collapses. I’m still putting the two under the same title here, beyond the obvious, because Kemp also talks quite a bit about deliberate democracy, the French Climate Assembly, and sortition. There’s overlap there between Spinks’ localism and becoming a person of place, and Kemp’s closer to the ground democracy.

The underlying issue is we’re being intentionally slowed down by big, unacceptable power structures like the fossil fuel industry. And having an enlightened dictator is not gonna change that. They too will just simply be captured by the fossil fuel industry and other big interests who have a vested stake in the fossil fuel based economy. The difference is that having [deliberate] democracy usually results in decisions where citizens realize that this is not in their best interest. And essentially they want to challenge power and they want to put society on a different track.

This one is a different kind of collapse, but it definitely would’t help dodging the other kind: Collapse of key Atlantic current could bring extreme drought to Europe for hundreds of years. “This is the first time that researchers have compared what would happen to Europe's summer precipitation under different climate scenarios if the Atlantic Meridional Overturning Circulation (AMOC) were to collapse.”


Futures, Fictions & Fabulations

  • Lecture with Simon Höher – Designing responsible futures. “What does it mean to design for a desirable future - and how does that translate into present-day choices? In this talk, Simon explores how our frames of thinking shape what we consider possible, and how finding glitches within them can open up new pathways.”
  • Buzzing Futures. “Buzzing Futures is an imaginative and interactive discussion game that invites players to explore the future of our ecosystems through the lens of pollinators. Developed by the Embassy of the Republic of Slovenia in the Nordics in collaboration with UN Global Pulse, the Secretary-General’s Innovation Lab, the game brings together creative storytelling, systems thinking, and play.”
  • Future Tense Fiction “Future Tense Fiction is a speculative fiction series that uses imagination to explore how science and technology will shape our future. It features short stories that are written by acclaimed authors across a wide range of styles.”

Algorithms, Automations & Augmentations

Built, Biosphere & Breakthroughs

  • Dutch farmers grow mushrooms in recycled egg carton towers. “A revolutionary mushroom cultivation method that employs vertical columns made from recycled egg cartons, entirely removing plastic from the growing process and optimizing sustainability and space use.”
  • Carbon Cell makes plastic-free alternative to polystyrene. “Looking almost identical to expanded polystyrene but in a moody black colour, the foam was made from biochar – the blackened remnants of biomass, in this case crop waste, that is created when it is heated through pyrolysis.”
  • A startup says it has found a hidden source of geothermal energy. The title is a bit misleading, but intriguing nonetheless. “Zanskar, which uses AI to find hidden geothermal resources deep underground, says that it has identified a new commercially viable site for a potential power plant. The discovery, the company claims, is the first of its kind made by the industry in decades.”

Asides

“Ambitious, thoughtful, constructive, and dissimilar to most others. I get a lot of value from Sentiers.”

If this resonates, please consider becoming a supporting member—it keeps this work independent.

Support Sentiers

Your Futures Thinking Observatory