--- title: "AI: The Weight of Weightless Things" description: "AI can draft contracts, automate decisions & accelerate work, but using it still demands knowledge, scrutiny and independent legal judgement." image: https://spoor.com/wp-content/uploads/2026/09/AI-scaled-e1790064826234.png canonical: https://spoor.com/ai-the-weight-of-weightless-things/ url: https://spoor.getmd.ai/ai-the-weight-of-weightless-things/.md last_converted: 2026-09-23T03:47:13.087Z --- ```json { "@context": "https://schema.org", "@graph": [ { "@type": "Article", "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#article", "isPartOf": { "@id": "https://spoor.com/ai-the-weight-of-weightless-things/" }, "author": { "name": "Lodewyk Cilliers", "@id": "https://spoor.com/#/schema/person/7c7741769609f890be17e826c8c88ba7" }, "headline": "The Weight of Weightless Things", "datePublished": "2026-09-22T09:36:38+00:00", "dateModified": "2026-09-22T09:39:05+00:00", "mainEntityOfPage": { "@id": "https://spoor.com/ai-the-weight-of-weightless-things/" }, "wordCount": 2174, "commentCount": 0, "publisher": { "@id": "https://spoor.com/#organization" }, "image": { "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#primaryimage" }, "thumbnailUrl": "https://spoor.com/wp-content/uploads/2026/09/AI-scaled-e1790064826234.png", "articleSection": [ "News Article" ], "inLanguage": "en-ZA", "potentialAction": [ { "@type": "CommentAction", "name": "Comment", "target": [ "https://spoor.com/ai-the-weight-of-weightless-things/#respond" ] } ] }, { "@type": "WebPage", "@id": "https://spoor.com/ai-the-weight-of-weightless-things/", "url": "https://spoor.com/ai-the-weight-of-weightless-things/", "name": "AI: The Weight of Weightless Things - Spoor & Fisher", "isPartOf": { "@id": "https://spoor.com/#website" }, "primaryImageOfPage": { "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#primaryimage" }, "image": { "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#primaryimage" }, "thumbnailUrl": "https://spoor.com/wp-content/uploads/2026/09/AI-scaled-e1790064826234.png", "datePublished": "2026-09-22T09:36:38+00:00", "dateModified": "2026-09-22T09:39:05+00:00", "description": "AI can draft contracts, automate decisions & accelerate work, but using it still demands knowledge, scrutiny and independent legal judgement.", "breadcrumb": { "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#breadcrumb" }, "inLanguage": "en-ZA", "potentialAction": [ { "@type": "ReadAction", "target": [ "https://spoor.com/ai-the-weight-of-weightless-things/" ] } ] }, { "@type": "ImageObject", "inLanguage": "en-ZA", "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#primaryimage", "url": "https://spoor.com/wp-content/uploads/2026/09/AI-scaled-e1790064826234.png", "contentUrl": "https://spoor.com/wp-content/uploads/2026/09/AI-scaled-e1790064826234.png", "width": 1200, "height": 550, "caption": "AI" }, { "@type": "BreadcrumbList", "@id": "https://spoor.com/ai-the-weight-of-weightless-things/#breadcrumb", "itemListElement": [ { "@type": "ListItem", "position": 1, "name": "Home", "item": "https://spoor.com/" }, { "@type": "ListItem", "position": 2, "name": "The Weight of Weightless Things" } ] }, { "@type": "WebSite", "@id": "https://spoor.com/#website", "url": "https://spoor.com/", "name": "Spoor & Fisher", "description": "", "publisher": { "@id": "https://spoor.com/#organization" }, "potentialAction": [ { "@type": "SearchAction", "target": { "@type": "EntryPoint", "urlTemplate": "https://spoor.com/?s={search_term_string}" }, "query-input": { "@type": "PropertyValueSpecification", "valueRequired": true, "valueName": "search_term_string" } } ], "inLanguage": "en-ZA" }, { "@type": "Organization", "@id": "https://spoor.com/#organization", "name": "Spoor & Fisher", "url": "https://spoor.com/", "logo": { "@type": "ImageObject", "inLanguage": "en-ZA", "@id": "https://spoor.com/#/schema/logo/image/", "url": "https://spoor.com/wp-content/uploads/2022/08/200210-SP-Logo_RGB-01-screen-scaled.jpg", "contentUrl": "https://spoor.com/wp-content/uploads/2022/08/200210-SP-Logo_RGB-01-screen-scaled.jpg", "width": 2560, "height": 813, "caption": "Spoor & Fisher" }, "image": { "@id": "https://spoor.com/#/schema/logo/image/" }, "sameAs": [ "https://www.facebook.com/spoorfisher/", "https://x.com/spoorfisher" ] }, { "@type": "Person", "@id": "https://spoor.com/#/schema/person/7c7741769609f890be17e826c8c88ba7", "name": "Lodewyk Cilliers", "image": { "@type": "ImageObject", "inLanguage": "en-ZA", "@id": "https://secure.gravatar.com/avatar/94e6d75b8199963bf82fa504b781281aed861108d3e8178c014b365a431a936a?s=96&d=mm&r=g", "url": "https://secure.gravatar.com/avatar/94e6d75b8199963bf82fa504b781281aed861108d3e8178c014b365a431a936a?s=96&d=mm&r=g", "contentUrl": "https://secure.gravatar.com/avatar/94e6d75b8199963bf82fa504b781281aed861108d3e8178c014b365a431a936a?s=96&d=mm&r=g", "caption": "Lodewyk Cilliers" } } ] } ``` [Home](https://spoor.com) › [Knowledge Centre](https://spoor.com/knowledge-centre/) › [Latest News and Insights/](https://spoor.com/knowledge-centre/latest-news-and-insights/) › The Weight of Weightless Things # The Weight of Weightless Things Lodewyk Cilliers - 22 September 2026 There are two kinds of people: those who read dystopian fiction to feel frightened, and those who read it to feel prepared. The worrying part is when the two start feeling like the same thing. I recently re-read The Day of the Triffids by John Wyndham, written seventy-five years ago but as unsettling as anything published last week. What gets me about the best dystopian fiction is never the catastrophe itself. It’s what comes after – the sudden revelation of how little any of us understands about the constituent parts of the world we depend on. In our hyper-specialised civilisation, almost no one comprehends the full system anymore. It was true in 1951 when Wyndham wrote Triffids. It is dramatically more true today. It happened gradually – until it didn’t. Ask ChatGPT to write you a sonnet and it will oblige in about two seconds. It feels like magic – pure thought, conjured from the ether. What you don’t see is the warehouse-sized building in Ohio humming with tens of thousands of processors, drawing enough power to light a small town, drinking millions of litres of water a day to keep itself from overheating. The gap between the effortless experience on your screen and the immense physical infrastructure behind it is a tension that has been quietly building inside human civilisation for centuries, and which artificial intelligence has made impossible to ignore. Here’s one way to think about it. Human civilisation runs on three systems, layered on top of one another. First, *communication*: the technologies through which we coordinate what we know. Second, *industry*: the technologies through which we convert energy into useful work. Third (and this one isn’t really a ‘system’ at all, but grant me the poetic licence) *thermodynamics*: the hard physical laws that govern how much energy is available and how much it costs to keep things organised. Communication and industry are things we built. Thermodynamics is the thing that was always there, indifferent to our ambitions. For most of history, what we could coordinate and what we could physically do grew more or less in step, and the physics underneath remained comfortably in the background. What makes the present moment so unusual is that artificial intelligence is pulling them apart. Start with *communication* – the system that determines how much complexity civilisation can coordinate. Every big leap in how humans talk to each other has made civilisation larger and more complicated. Grunts and gestures got us through hunting in small bands. Spoken language gave us the ability to name, to plan, and to teach – knowledge survived a generation and a larger tribe. Writing made knowledge durable and portable. An idea could now survive its author and travel further than any voice could carry it, in so doing supporting more complex societies. The printing press blew the doors off: once you could copy a book cheaply, you had mass literacy, the scientific revolution, and, eventually, factories. Radio, television, and the internet compressed the whole planet into a single nervous system. The pattern is consistent. Each communication revolution let more people coordinate around shared knowledge. More coordination meant more complexity – denser networks of institutions, trade routes, legal systems, supply chains, and mutual dependencies. Civilisations grow not only because they have resources, but also because they figure out how to organise those resources at an increased scale. AI looks, at first, like the next step in that story. But something feels different this time. Every previous communication revolution helped humans communicate better. AI, however, doesn’t just help us talk – it also does some of the thinking. It interprets, synthesises, translates, plans, and makes decisions. Information is no longer just carried by machines. It is generated, filtered, and acted on by them. Communication seems to be shifting from expression to computation. And that may mean the current communication system is no longer just supporting civilisation from the sidelines. It is starting to become a participant in its own right. Now *industry* – the system that determines how much physical force civilisation can deploy. If communication set the ceiling on complexity, industry set the ceiling on power. For most of human history, the ceiling was quite low. Everything ran on muscle – human and animal – plus whatever you could coax out of wind and water. Then came steam, and suddenly a machine could do the work of a hundred men. Electricity and oil made factories faster, bigger, and more interconnected. Computing automated calculation and logistics, and made machines more efficient and intelligent. At each stage, civilisation outsourced another piece of human capability: first muscles, then energy conversion, then the tedious parts of thinking. The trajectory is consistent: each revolution outsourced something higher up the chain of human capability than the last. AI extends that trajectory into the one domain that previously resisted externalisation: cognition itself. And here’s what makes this moment feel different. Communication and industry used to be recognisably separate things. The printing press was not the steam engine. The telegraph was not the assembly line. One system coordinated civilisation; the other powered it. Software began blurring that line decades ago. A spreadsheet communicates and produces at the same time. A search engine is simultaneously a knowledge system and a commercial platform. But AI may be accelerating the blurring to a point where the distinction becomes difficult to maintain. Think about what a large language model actually is. Communication system – you talk to it, it talks back. Knowledge store – it has absorbed more text than any human could read in a thousand lifetimes. Productive tool – it writes code, drafts contracts, designs circuits. Coordination mechanism – it routes queries, manages workflows, automates decisions. Previous communication technologies primarily transmitted what humans want them to transmit. Previous industrial technologies primarily amplified human labour and skill. AI increasingly does both at once. This may be why the “Fourth Industrial Revolution” never quite felt like the right label. Steam was a thing. Electricity was a thing. AI isn’t really a thing in the same way – it’s more like a fusion of computation, data, networks, automation, and cognition into a single integrated system. Its centre of gravity isn’t mechanical power. It doesn’t run on horsepower. It runs on pattern recognition, language, and logic. And that changes what becomes possible. Until now, the limit on how complicated civilisation could get was the human mind. You could only build systems as intricate as the people running them could understand. AI is raising that bar dramatically – and that’s not hype, it’s real. Problems that would have occupied entire careers are becoming solvable in a fraction of the time. But a structural consequence may follow. If the bottleneck was always brainpower, and AI loosens it, what stops civilisation from attempting things that are cognitively possible but physically unsustainable? The answer is the third layer: *the laws of* *thermodynamics* – the hard limits within which the other two must operate. Two laws matter here. The First Law says energy cannot be created or destroyed, only converted – there’s a fixed amount to work with, and every use has a cost. Much of the current conversation about AI and energy is about the First Law: computation requires power, and power isn’t free. Data centres are projected to consume more electricity than Japan by 2026. That’s important, and it’s well understood. A quick look at geopolitical instabilities will confirm – the race for energy is well and truly on. But the Second Law may matter more for what comes next. The Second Law says that every energy conversion wastes some of it, scattering energy into less useful forms. Left alone, systems tend toward disorder, and maintaining order requires a continuous input of energy. Stop feeding energy in, and things fall apart – a truth that every homeowner, mechanic, and gardener discovers independently. Scientists call this tendency entropy. The more complex the system, the more energy it takes to keep things organised. Civilisation is an ongoing argument with disorder. We build extraordinary complexity, and we sustain it only through continuous effort and energy. The moment the input stops, the drift toward chaos resumes. Order is not owned – it’s rented. I want to borrow the Second Law idea, not as physics, but as a metaphor. Because something very similar applies to knowledge. Order requires effort to maintain – and that effort isn’t always measured in kilowatt-hours. Sometimes it’s measured in attention, understanding, and intellectual discipline. Every previous communication revolution demanded that humans learn something new. Writing required literacy. Print required education. The internet required digital competence. Each time, the price of admission was intellectual effort. The effort to participate meaningfully in the new regime and to remain capable of checking, questioning, and understanding the systems you depended upon. AI is different. It lowers the barrier so dramatically that you can use extraordinarily powerful systems without understanding them at all. Draft a contract, diagnose a symptom, design a structure – all possible without possessing much of the underlying knowledge those tasks traditionally required. The cognitive admission ticket is diminished, and with it, potentially, the compulsion to learn. The electricity bill for AI is visible and measurable, but the intellectual bill is not. It’s the second-law cost of the revolution – the effort required to stop our own understanding of an increasingly complex world from quietly degrading into chaos. None of this should be mistaken for skepticism. AI is remarkable, and its potential is real. But so is the cost of using it carelessly, and that cost isn’t measured in kilowatt-hours. It’s measured in the slow, quiet erosion of the understanding you need to know when something has gone wrong. The question isn’t whether to use AI. It’s whether we’re prepared to invest the effort that using it properly requires. AI does an extraordinary amount of the heavy lifting. But it does not exempt us from the effort to understand. The more powerful the tool, the more consequential the failure to understand how it works, what it assumes, and where its limits lie. The electricity is only half the equation. The other half is the intellectual energy to keep ourselves literate in the fundamentals – to check, to question, to recognise when something has gone wrong. That second kind of energy is not optional. ***RELATED: [The Toggle That Launched a Thousand Claims](https://spoor.com/ai-tools-ownership-and-liability/)*** What this means in practice? Using AI well is becoming a core skill. However, that skill is not about outsourcing thought. It’s about redirecting it. Used correctly, AI should free up time for exactly the kind of high-level analytical thinking that matters most: framing the problem, interrogating assumptions, weighing competing interpretations, and exercising judgement. The hard yards of reading, collating, and summarising can increasingly be delegated. The thinking cannot. Think of AI as a sparring partner, not a delegate. Put in the work at both ends – clear direction and well-considered prompts at the front, critical judgement and testing at the back – and let AI do the heavy processing in the middle. Part of the skill is learning to engage with the tool itself: knowing where it misses nuance, where it hallucinates confidence, and doing that without letting your own foundational skills diminish. This is not a binary choice. Independent thought and AI assistance need to live in parallel, each strengthening the other rather than one replacing the other. A warning, though: this balance is hardest for those still building foundational knowledge. If you haven’t yet developed the instincts to know what “good” looks like, AI can feel like a shortcut that works – right up until it doesn’t. For juniors, students, and anyone early in their discipline, the integration of AI into learning needs to be deliberate and clear about what the tool can’t teach you. The entropy tax is steepest for those with the least experience to draw on. I don’t have a grand conclusion. The honest truth is that nobody does. Not yet, and probably not for a while. We built a machine that thinks faster than we do and eats more electricity than we’d like to admit, and now we’re all standing around it like ancient villagers around a recently tamed fire, mesmerised by the warmth and trying not to think too hard about what happens when it gets out of the pit. The sensible approach is probably to learn what the thing actually does, stay curious about what it can’t do, and do the work to keep your own thinking honed. Civilisation has survived every previous revolution by muddling through with generally adequate understanding. The question is whether “generally adequate” still cuts it when the tools are this powerful and the stakes are this high. I suspect the answer is no. But I also suspect we’ll try it anyway – and will figure it out as we go along.