AI gets stuck in infinite loops ⊗ A real genius club ⊗ The metabolic logic of AI slop

💡 Reading a friend’s sporadic newsletter made me wonder something. Would interesting writers be better off contributing to a regular, and quite possibly larger, weekly publication rather than posting sporadically on their own? In other words, perhaps they (you?) would be interested in writing here as columnists or editor-at-large? Hit me up for thoughts or interest.

Why AI gets stuck in infinite loops — but conscious minds don’t

Anil Seth argues that AI systems can get trapped in infinite loops because, unlike conscious beings, they lack an intrinsic connection to physical time and entropy. According to his thinking and research, human intelligence is deeply tied to consciousness, which is itself linked to the continuous flow of time and the imperative to survive in a world governed by the second law of thermodynamics (in short, entropy). Computers, operating through timeless algorithms, cannot fully replicate this time-embedded, adaptive intelligence, which, in Seth’s opinion, makes genuine AI consciousness unlikely in the near future.

It’s quite an intriguing idea. However, although I can’t really make a solid argument for it, it seems like just an interpretation of how LLMs work vs time, I’m not convinced this distinction necessarily holds the road. On the other hand, you can also take it as a different version of the embodiment argument, where multiple people believe that true intelligence and consciousness can only be achieved through physical interaction with the world.

Also, I’m going to have to revisit Venkatesh Rao’s Superhistory, not superintelligence (from way back in issue No.173) where he “proposes that Artificial Intelligence is actually Artificial Time, that it compresses decades, centuries of learning in hours and minutes and makes it available to us as augmentation. I.e. AlphaGo played millions of games in a short time.” Not the same thing, but those two aspects of time as it relates to LLMs is something worth thinking about further.

This disregard for time is costly. While computation, as Turing formalized it, exists out of time, computers themselves do not. The insatiable energy appetite of today’s AI is largely due to the demands of error correction — the need to keep ones as ones and zeros as zeros — since even the dead sand of silicon cannot escape the tendrils of entropy. Pretending that time doesn’t exist is an expensive business. […]

Unlike computers, we are beings in time — embodied, embedded, and entimed in our worlds. We can never be caught in infinite loops because we never exist out of time. The constant, time-pressured imperative to minimize statistical surprise and maintain physiological viability is, for living creatures like us, the ultimate relevance filter, the reason we almost always find a way through. […]

According to some prominent ideas in neuroscience, notably the free energy principle, neural circuits continuously work to minimize the statistical surprise of sensory inputs, to reduce the entropy of these inputs, and in doing so, impose a kind of predictive control on the body’s physiological condition. […]

That type of intelligence, I believe, is intimately and perhaps necessarily tied up, not only with time, but also with consciousness. At least, if you conceptualize consciousness the way I do — as a process deeply linked to the drive to survive.

Inside a real genius club

This is a 42 minute documentary, hosted by Kmele Foster, about the Santa Fe Institute. Don’t make the mistake I made, listen to it in one go, scenius meets genius meets thinking retreat meets big ideas.

Founded in 1984 by Nobel laureates and other distinguished scholars, SFI emerged from a conviction that traditional academic boundaries were preventing scientists from tackling life’s messiest, most important questions. The institute studies complex systems—ant colonies, economies, cities—searching for underlying rules that govern how simple interactions create staggering complexity. David Krakauer, the Oxford-trained evolutionary biologist who leads SFI, describes it as a “monastery in the mountains,” a place where researchers can test ideas within a supportive community before taking them to the wider world. Below, in lieu of quotes, summaries of two of my favourite parts.

When asked about life’s big questions, theoretical physicist Geoffrey West offers a surprisingly direct answer. He identifies two fundamental purposes for human existence. First is love in its broadest sense—the capacity to care, which he sees as uniquely human. Nothing else in the universe, he argues, possesses morals or ethics; trees and animals don’t care, but humans do, despite our capacity for evil. Second, and perhaps more fundamental, is understanding. West views humans as the mechanism through which the universe becomes self-aware. Walking through a forest, he experiences something almost spiritual: while the trees embody order without knowing it, humans have been “blessed" with the ability to recognize and comprehend that order. We are, in his formulation, “the thoughts of the Universe.”

During a group lunch, unnamed researchers explore identity and purpose through surprising metaphors. One participant describes himself as a mushroom—merely the visible fruiting body of a vast underground network of information flowing from parents, ancestors, bacteria, and daily interactions. This leads to a debate about whether the universe is truly purposeless. While popular science often declares it so, one researcher points out the obvious contradiction: the universe clearly generates purpose since humans demonstrably have purposes every time they eat or act.

Eating the future: the metabolic logic of AI slop

Kate Crawford riffs on a variation of the metabolic rift, “the systemic disruption of ecological and metabolic processes by capitalist production.” She explores how generative AI transforms human culture into a vast resource of data, producing a flood of synthetic, hyper-realistic images and content that feeds back into AI systems. This process creates a new metabolic rift, disrupting ecological, cognitive, and model stability, driven by enormous energy and water consumption that strains planetary resources. To move beyond this destructive cycle would require fundamental changes in technology, ownership, and cultural practices, to prioritise ecological regeneration and human flourishing.

Nate Hagens uses the concept of the Carbon Pulse, “a one time massive consumption of fossil hydrocarbons at a pace millions of times faster than they were created.” Reading Crawford, I kept wondering if we are in an AI Pulse, a one time massive consumption of data at a pace millions of times faster than they were created. I was also thinking of Charlie Strauss’ post on mistaking a transient state for a permanent one. Is LLM training a transient state before we run out of data, burst a bubble, and/or transition to much less energy-intensive smaller models? On the flip side, as ever, is our old friend William Stanley Jevons and his paradox. (“In addition to reducing the amount needed for a given use, improved efficiency also lowers the relative cost of using a resource, which increases the quantity demanded.“) No conclusion here, just lots of questions and few certainties.

The AI slop aesthetic is more than a technical artifact or cultural curiosity. It’s the consumer-facing product of computational capitalism—created by transforming human culture into a standing reserve of data. Each uncanny image, synthetic influencer, and AI-generated article is part of a much wider transformation. Slop is waste, but it’s also fuel. […]

Metabolic images are visual material made by consuming billions of other images to be broken down and absorbed by AI models via a gargantuan infrastructure that absorbs energy and water, and excretes media, atmospheric carbon, and other pollutants. […]

The metabolic rift between humans and AI threatens multiple forms of failure: model collapse, ecological collapse, and cognitive collapse. The current approach to AI cannot sustain itself indefinitely. The question becomes not whether current AI slop economies will destroy themselves, but when. […]

Such alternatives would demand the fundamental transformation of existing technical and economic arrangements—alternative energy systems, cooperative ownership models, and cultural protocols that resist algorithmic ingestion. This requires not a nostalgic return to pre-digital cultural forms, but practices adequate to our ecological and social moment.

§ It’s time to work different. Great list of shifts to “toss the hype, index on humans.” “Prompts are not the magic key to AI—most folks I’m talking today have come to the same conclusion. Prompts are just the syntax while literacy is the fluency. Literacy means you understand context, nuance, and when to ignore the machine altogether. Without literacy, prompts are just the modern version of shouting at Clippy and hoping it understands you and will shield you from imposter syndrome.”

§ SPOS #1000 – Patrick Tanguay On paths to better thinking. If reading my thoughts here isn’t enough (ah!), Mitch Joel made me the honour of being the 1000th (!!!) guest on his podcast. Fun conversation on a bunch of topics we are both thinking about. “This is an episode for curious minds, digital gardeners and anyone searching for clarity amid chaos.”

Futures, Fictions & Fabulations

• 2025 Strategic Foresight Report – European Commission. “Presents "Resilience 2.0", an approach to ensuring the EU thrives in turbulent times through 2040 and beyond. It builds on the recent European Preparedness Strategy and recognises the need to scan proactively not only for emerging risks but also for future opportunities, and to consider unfamiliar or even hard-to-imagine scenarios.”
• Frontiers 2025: The Weight of Time – Facing a new age of challenges for people and ecosystems. “The Frontiers Report spotlights emerging environmental issues before they escalate into global or regional crises. While these issues may currently appear localised or small-scale, early intervention is critical to prevent them from becoming widespread challenges.”

Algorithms, Automations & Augmentations

• The hot air factory. Fun project by the creative folks at oio. “We all use AI prompts every day, and we know they consume lots of energy. But do we really know how much? We didn’t! So we built a different kind of AI. It’s a tiny box with a pinwheel. We call it the Hot Air Factory.”
• I hate my Friend. “The chatbot-enabled Friend necklace eavesdrops on your life and provides a running commentary that’s snarky and unhelpful. Worse, it can also make the people around you uneasy.”

Built, Biosphere & Breakthroughs

• The first evidence of a take-off in solar in Africa “There has been a major pick-up in solar panel imports into Africa over the last 12 months – a shift that is likely to impact almost every country on the continent.”
• Finnish city inaugurates 1 MW/100 MWh sand battery. “The concept is simplicity itself. Make a really big pile of sand. Heat it with excess renewable electricity to around 500°C (932°F), then use that heat later to heat homes, factories, even swimming pools. They say the sand can stay hot for 3 months or more.”

Asides

• Dennis Lehtonen documents a pair of immense icebergs paying a visit to a small Greenland village. Yikes! There’s an art project I’d like to see; replacing icebergs with cruise ships and doing the opposite with pics of Venice and elswhere.
• I planted a forest four years ago. Well done sir, well done. “In 2021 I planted a tree a minute, for 24 hours, on my mates farm. It was freakin hard work, but also one of the coolest, most rewarding days I’ve ever had. I made a film about the project and promised folks I’d return every two years to show off the plot and see how the trees and bushes are going. This was a special day because I really felt like the project had landed. I had a cup of tea in the new forest, from water boiled on a fire made from the forest itself. It’s perhaps the most profound cup of tea I’ve ever had.”