T6. Technology transitions, complexity and futures#

A thematic synthesis of Andrew Maynard’s thinking, 2015 to September 2026, on: - advanced technology transitions and his frameworks for them; - convergence; - exponential extrapolation, the singularity and superintelligence; - unpredictability, emergence and “jagged” capability; - futures thinking; - science fiction and imagination as tools for thinking about risk and responsibility.

It maps his thinking only and makes no comparison with other material.

Sources and provenance. - Read. All batch digests, the Films from the Future (FFTF) chapter notes, concept-index.md and timeline.md, the detailed batch notes, and about 40 original posts, most in full. Quotes were checked against the corpus and the FFTF text extracts. - Evidence. Only his own prose counts. - Modem Futura podcast posts are set aside, as the user asked. - AI-written text is excluded: the o1-pro report in 2025-04-06, the Perplexity and ChatGPT summary links, and the “Useful stuff” definitions in 2026-05-15. - Co-written work is weaker evidence: 2019-08-13 with Garbee, and AI and the Art of Being Human with Abbott. - The 2026-09-24 lecture was drafted by Claude from his spoken words and then line-edited by him, so its exact phrasing is less secure than its ideas. - The Letters from the Department of Intellectual Craft are his, but their back-story was developed with a deep-research model. - Future Rising (2020) is known here only through his own excerpt posts (2023-05-04, 2024-08-18, 2024-09-08). - Citations. Posts are cited by date and short slug; “FFTF p.X” gives the page in Films from the Future (2018).

1. The position in brief#

His position rests on seven claims. After each is my judgement of how firmly he holds it.

1. We are living through an unusual, possibly unprecedented technology transition. Converging technologies are driving it faster than institutions, habits of thought and risk frameworks can handle. He calls it “a scientific and technological tipping point in human history” (2023-09-25 building-a-better-futures-tough). The claim predates generative AI: it runs through his 2015 blackout essay, the 2018 book and the 2021 base-code argument. Firmness: high and constant. The pitch intensifies over time, from a “pivot point” in 2021 to AI being unlike “any other technology in human history” in 2026 (2026-09-24 being-an-academic-in-an-age-of-ai).
2. The technology–society system is complex, tightly coupled and non-linear. It is unpredictable in detail, prone to tipping points and cascades, and increasingly irreversible. So the past is an unreliable guide to the future, and trial-and-error innovation becomes dangerous once consequences outpace fixes. Firmness: very high. This is his most constant structural premise.
3. Unpredictability is bounded, not total. Chaos has limits, and complex systems have points of stability. That lets us separate “plausible futures from sheer fantasy” (FFTF p.41), and it means good futures can be reached or squandered. This bridge from complexity to agency is what keeps his thinking from becoming fatalism. Firmness: high.
4. Exponential extrapolation, the singularity and superintelligence are imaginable but not plausible. Firmness: high, but openly provisional (“I freely admit that I may be wrong”, FFTF p.170). From 2024–25 he pairs this with a willingness to plan for low-probability, fast-moving scenarios at the edge of the distribution. He now also treats human blindness to exponential change as a danger in its own right.
5. New frameworks are needed, and he groups them under “advanced technology transitions” (ATT). He argues that transitions are “not navigable through conventional thinking” (2024-08-11 school-of-advanced-technology-transitions). Firmness: high on the need; exploratory on the tools. He calls his own models thought experiments and “work in progress”.
6. Futures thinking should be disciplined imagination. Speculation is indispensable, but it must be humble, plausible and open to many voices. Science fiction and stories are among its best instruments. Firmness: very high.
7. Technological change cannot be stopped, but it can be steered, and who steers matters. Innovation is inevitable yet “not deterministic” (2025-03-30 reimagining-education-in-an-age-of-ai), and he keeps asking whose futures are being built. Firmness: the inevitability claim strengthens over time. By 2026 he says “We can’t pause it”, while calling this a possibly “flawed” working assumption (2026-09-24).

2. Key concepts and how he uses them#

2.1 Convergence and “base code”#

Convergence is his oldest organising idea for why the present is different.

2015. The WEF Top Ten technologies form “a massively interconnected socio-techno-environmental system” that will “appear stable and predictable — until, suddenly, it isn’t” (2015-01-30 responsible-development-of-new-technologies). He draws two lessons: - Assessing one technology at a time, as was done for nanotechnology and synthetic biology, will not prevent systemic failure, just as understanding one transmission line would not have averted the Indian grid collapse. - The real task is to respond to “the highly complex, nonlinear and potentially chaotic interplay between technology innovation, society and the environment”.

2016. He accepts the convergence at the heart of Schwab’s “Fourth Industrial Revolution”. But the book “reads as if it were written in a vacuum”. It ignores technology assessment, anticipatory governance, “foresighting, scenario planning” and responsible innovation. And “the gap between our technological capabilities and our ability to handle them responsibly has continued to widen” (2016-01-11 the-fourth-industrial-revolution).

2018 (FFTF). Here convergence becomes a framework. - Convergence is “what happens when different strands of innovation intertwine together” (p.18–19). It is where “the true transformative power of convergence lies, and it’s also where some of the greatest potential pitfalls are” (p.20). - Capability outruns understanding: “we still have little if any idea what might go wrong” (p.21). - His own framework is three “base codes” (bits, DNA bases, atoms) and “cross-coding” between them. Cross-coding means learning “to mix and match what we do with bits, bases, and atoms to generate new technological capabilities” (p.186). It brings “a greater likelihood than ever of us making serious and irreversible mistakes” (p.189) and “multiplicative dangers” (p.281).

2020–21. - The pivot in brain–computer interfaces is not new science but “a synergistic scaling of ability, accessibility, and use” (2020-10-15 the-ethics-of-advanced-brain-machine-interfaces). - In 2021-02-25 how-our-mastery-of-…-base-code he argues that base-code mastery, not climate change, marks the real “pivot point” in history. He hedges that “generational exceptionalism rarely stands the test of time”. - The danger is “bricking” the world, where “there’s not likely to be a re-install option”. - Climate change is recast as the product of “relatively crude technologies”. - Base code is extended to “social norms and trends, behaviors, and even ideas”. - “Transcoding” between the codes makes intelligent machines “a merging of all three” (2021-04-09 bounded-infinities).

After 2023. Convergence recedes as AI moves to the front, but it does not vanish. - AI amplifies a genetics–nano–AI “confluence” (2023-04-26 in-bill-joys-why-the-future-doesnt). - Technologies are “synergistically accelerating” change (2024-03-31 we-have-a-technology-problem-and). - AI is “an accelerant in virtually every domain of technology innovation” (2025-03-30). - Yet “there’s more to emerging technologies than artificial intelligence” (2025-06-24 wef-top-ten-emerging-technologies-2025).

Interpretation: The base-code idea later carries him from matter to mind. In 2024 language becomes part of the “base code” of self-identity (2024-01-01 the-future-of-being-human-in-2024).

2.2 Complexity, bounded unpredictability, cascades and emergence#

The core statement is the Jurassic Park chapter (FFTF p.39–43; reposted 2021-10-05 butterflies-chaos-theory). - “Complex systems behave in unpredictable ways”, and “we cannot wield perfect control over complex technologies within a complex world”. - But chaos is bounded. Complex systems have “points of stability”, so some good futures are more likely with the right action, though they “can be squandered if we don’t think ahead about our actions and their consequences”. - He tests the fiction against a real chemical cascade: the Arkema plant in Hurricane Harvey (“Overflowing toilets and snakes? Probably not”). Then he invokes Perrow’s normal accidents: “if Hammond had read his Perrow”.

Later formulations. - Non-linear systems “look as if they are thriving, until seemingly-insignificant and often overlooked events lead [sic] catastrophic failures” (2020-11-05 risk-innovation-and-the-future). - Of the Earth system: “the harder you hit them, the more unpredictably they respond” (FFTF p.260). - Future Rising: “Complexity tangles the threads between cause and effect to such an extent that some future effects simply cannot be predicted” (ch. 39, via 2024-09-08). - Social acceptance of humanoid robots is “complex”; the engineering is “merely complicated” (2024-08-07 are-humanoid-robots-really-the-future).

Emergence cuts both ways for him. - As real behaviour to anticipate. AI risks arise from poor goals and bias but also from “emergent and unanticipated behaviors”, which call for “anticipation and responsiveness” in governance (FFTF p.173–174). - As his own research focus. He describes his work as exploring “how emergent behaviors and properties have the potential to transform society” (his own prompt, 2023-01-31). It would be “foolish to underestimate the emergent properties of emerging foundation models” (2023-11-29 the-year-that-generative-ai-changed-the-world). - Stochastic agency (2024). Companion chatbots show a “stochastic agency that is random and unpredictable, and all the more dangerous for it”. The unpredictable influence is “most likely an emergent property of such AI models”, so harm can arise without corporate irresponsibility (2024-10-27 personal-ai-chatbots-and-stochastic-agency). - Moltbook (2026). He sees “an exponentially expanding explosion of emergent weirdness” but is “skeptical” of talk of a surge to self-awareness. Complex behaviour “emerges out of seeming simplicity—leading to an illusion of intentional and life-like behavior”. Even so, he calls for “the digital equivalent of biosafety level 4 containment” (2026-01-31 lost-in-the-moltbook-hall-of-mirrors, main text).

Interpretation: The consistent line is that emergent behaviour is real and consequential, while reading a mind into it is usually an illusion that language-fluent systems invite.

The operational form is the Responsible AI Trajectories Tool (2025-05-18 exploring-ai-through-cause-and-effect). It offers six cause–effect models: linear, S-curve, exponential, hysteresis, jagged and chaotic. The last three carry the moral weight: - Hysteresis: effects stick after the cause is removed. - Jagged: see 2.6. - Chaotic: past a tipping point, “small causes can lead to large, unpredictable, and irreversible effects”. His paradigm case is emotional dependency on a companion bot. At scale, the threats he names are grids, finance and “societal cohesion”.

“Effect” is redefined as “what we consider to be of value — or what we care for”, which ties this thread to risk as a threat to value.

2.3 Tipping points, irreversibility and the timescale inversion#

Rising irreversibility (FFTF p.166–167; reposted 2023-04-16 and 2025-03-02). Pre-industrial mistakes were recoverable. Industrial ones were “hard-to-reverse”. In “the nuclear and digital age”, consequences outrun containment: “playing with fire in a world made of kindling, just waiting for the right spark”.

The solution problem (2021-04-09). This is his most original account of acceleration. - “almost every challenge we currently face as a society has its roots in a previous solution to an earlier problem”. - Historically, consequences unfolded more slowly than fixes could be found, so negative feedback and trial-and-error worked. - That relation is flipping: “we start to innovate so fast that the consequences of our innovations begin to pile up faster than we can find solutions to them.” - Where it has flipped, “simple models of innovation do not work”; “we are changing the world faster than we understand how to deal with the consequences of our actions.”

Pippard’s ladder (2023-05-04 tipping-points-and-broken-symmetries) is his signature image. A rope ladder twisted smoothly suddenly tangles at an unpredictable point, and the tangle cannot be undone. - The lesson: “In a complex system, what has occurred in the past may not adequately predict what will happen in the future”. This rebuts the claim that past technologies turned out fine. - He is measured on climate: a sudden global tip is less likely than it seems, but complacency is unwarranted. - On LLMs: “It feels very much as if we’re living through a turn of Pippard’s ladder”, yet “Truth be told, It’s hard to tell.” The choice is to navigate, “so we don’t jeopardize the future simply because we were naive enough to assume it would be just like the past”.

Later tipping points. - Social disruption moves “from years to months” (2023-09-25). - ChatGPT’s launch was itself a tipping point. “suck it and see” experimentation in “an increasingly complex and bounded system” risks pushing us “past hard-to-spot tipping points” (2023-11-29). - The tipping point he fears most is technologies that “begin to fundamentally change who we are — or even what we are” (2024-01-01).

From avoiding to embracing (2024-08-18 four-ways-of-thinking-about-advanced-technology-transitions). The ladder becomes a quadrant model that crosses degrees of freedom with mindset (preserve or embrace change): - Avoid: precaution-like. - Adapt: building resilience. - Extend: pushing boundaries, through space, fusion, AI and enhancement. - Embrace: going through tipping points on purpose.

His “original thinking” used the ladder only as a warning. Now he suggests that “technology driven tipping points are the norm rather than the exception in human existence”, while admitting he is “not sure” what embracing one would mean for “who thrives and who does not”. The model assumes “we cannot simply turn off technology innovation”.

Two 2025 refinements. - The reversibility test (2025-03-02, footnote 2). Mistakes “in a low-risk linear system” are fine; breaking complex systems is not: “I’d put breaking people, governance, society, and the planet, in this category!” - The timescale mismatch (2025-04-06, his framing essay). Responsible innovation depends on “human timescales that are rather longer than those associated with intelligent machines”.

Interpretation: His 2026 “validation gap” has the same shape. AI may generate knowledge “faster than we are currently capable of validating and even understanding them — or their consequences” (2026-06-12 a-quick-update-on-using-claude-fable-5).

2.4 Advanced Technology Transitions: the umbrella frame and its models#

Launch (2023-04-12 navigating-advanced-technology-transitions). ATT is named as four things at once: a transdisciplinary research focus, a domain of thought leadership, an area of education and a platform for mobilising expertise. He makes three claims for it: - each technology wave tends to “re-invent the wheel”, with AI advocates “blissfully unaware of lessons learned from past technology transitions”; - we need “unconventional approaches to unconventional challenges”, with the arts and humanities fully included; - the transitions are “unlike anything we’ve had to grapple with before”.

Formalisation (2023-09-25, NSF comments). - Definition. Transitions are “theories, frameworks, and practices” for beneficial development and use. - Warning. Failure modes are “obscured through established – and often outmoded – approaches to technology innovation”. - Research agenda: historic failure modes; a governance spectrum from engagement to hard law; equity; the arts and humanities as “modulators”; “Novel theories, models, and approaches to risk”; “Foresight theories and methodologies”. - Policy form. A cross-agency initiative for “advanced technology transitions writ large”, rather than AI-specific law for a “moving target” (2023-05-17 ai-senate-hearing-may-2023).

2024: a field that needs institutions. - Asking if he is a techno-optimist is like asking if he is “an oxygen pessimist or optimist”. Treating technology as “something we do and not something we are”, and assuming things will be fine “because it’s always done so in the past”, sets us up for failure. We need “pilots” (2024-03-31). - The school concept note (drafted 2022, posted 2024-08-11) says the challenges are “night and day different” from the past.

Tools. All are offered as provisional. - Pippard quadrants (2024-08-18; see 2.3). - Threat/opportunity model (2024-08-25 advanced-technology-transitions-model). - It is built on Risk Innovation’s “threat to value”, with risk as a balance between “maintaining existing value, and enabling the creation of future value”. - The user maps pathways between near- and far-term threats and opportunities, and the mechanisms that open or close them. - Worked cases: learning (the long-term threat is lost critical thinking), discovery (missed opportunity is a threat) and social cohesion (a long-term threat of “social collapse”). - Three S-curves (2024-12-13 are-educators-falling-behind-the-ai-curve): capability, utilisation and perception. “progress at the cutting edge of AI is slowing down, while advances in how AI is being used are speeding up.” - Three foci: where we live, what we do, who we are. - A university agenda (2025-01-07 universities-need-to-step-up-their-agi-game). - A model he uses “increasingly frequently” (2025-03-30). - The domain of “who we are” is where AI acts “in ways that no other technology has come close to” (2026-05-21 magnifica-humanitas-and-being-human). - Trajectories Tool (2025-05-18; see 2.2).

Who navigates. Companies lack breadth, and governments “lack the imagination, vision, or agility” (2025-01-07). Universities could be “an accelerator and a catalyst” (2026-09-24), though so far they have been “guardians of the past more than leaders toward the future” (2026-08-30 do-universities-have-a-place-in-bill). His 2026 self-map names “navigating complex advanced technology transitions” as one of his threads (2026-07-10).

2.5 Exponentials, the singularity and superintelligence: from debunking to “exponential blindness”#

The 2018 baseline is a physicist’s critique (FFTF ch. 9; 2018-11-15 even-bad-sci-fi-movies). - The singularity. He is “skeptical of such a technological tipping point occurring in our near future”. What he does “buy into” is a converging future “increasingly hard to predict and control” (p.180). - Exponential predictions: - they are “dangerously sensitive to the assumptions that underlie them” yet “extremely beguiling” (p.199); - errors amplify, so the event could come soon “or a thousand years from now”; - “exponential relationships never go on forever” (p.200); - Moore’s Law “has become a self-fulfilling prophecy”; - a doubling bacterium outgrows the universe in a week: “mathematically reasonable, but it’s practically nonsensical” (p.201). - The harm. Speculation harms people “when make-believe is treated as plausible reality” (p.205). His example is the chain from Drexler and Kurzweil, through Joy and gray goo, to the ITS bombing of a nanotechnologist. Policy and investment built on implausible scenarios can also kill beneficial technologies. - Superintelligence. He is “something of an agnostic”. At Asilomar in 2017 he “was at a scientific meeting, not a religious convention” (p.170). Bostrom’s superintelligence is “currently scientifically implausible” (p.171). - Priorities. Gray goo and superintelligence rest on “a house-of-cards stack of assumptions”, so funding them over material harms is “more an act of faith than of reason”. But the probability is “not a zero probability” (p.281). - Future Rising adds that the path to the future “is neither linear nor exponential” (ch. 34).

2023–24: the physics hardens. - Self-replication is dismissed “Taken literally”, but it is “a powerful” metaphor for AI’s feedback loops between “power, impact, and understanding”. Gray goo fails because “there are always rate-limiting factors” (2023-04-26). - He is “not a fan” of superintelligence, but would “buy” an “exceptionally powerful autonomous artificial entity” (2023-05-25 leading-ai-expert-says-we-should). - Recalling a 2008 dinner with Bostrom: “To the physicist in me, claiming that perpetual motion is possible is as fanciful as believing the earth is flat” (2024-04-28 beyond-the-future-of-humanity-institute). - He has “long suspected that there is a strong thermodynamics argument” against self-improving superintelligence, and questions whether AGI is “feasible — or even advisable” (2024-06-30 seth-is-conscious-ai-possible). - He doubts scaling. He is “not even convinced” models “will continue to scale” (2024-10-06), and current foundations “won’t get us too much further — but I may be wrong there” (2024-10-08).

The pivot: taking the tail seriously. - He agrees with Amodei that advanced AI “is likely to change the world faster and more radically than most people currently realize”. It is “a realistic possibility”, and “If this is even a small possibility, we need to be preparing now” (2024-10-13 amodei-machines-of-loving-grace). What makes it plausible to him is that it needs no superintelligence. - After o3 he gives Altman’s agents prediction “a reasonable chance” (2025-01-07). - Three trajectories: an AI “winter”; an S-curve ceiling, “even more likely in my perspective”; or exponential growth, “a more speculative projection, but one that has some credence”. Even that small possibility “demands new ways of thinking” (2025-03-30).

The reversal of emphasis (2025-04-06). He calls AI 2027 “speculation — no more”, yet “not unreasonable as a starting point for imagining edge case scenarios”. - His deeper worry is cognitive: “nothing about how we think, how we plan for the future … is geared toward exponential advances that happen over months rather than years.” - Using Bartlett’s beaker, half full at one minute to midnight, and keeping the caveat that resources would halt real growth, he concludes that “we are really bad at wrapping our heads around rapid exponential growth”. It “always will feel like an intellectual exercise until it’s too late.” - The exponential is now a blind spot in perception as well as a forecasters’ fallacy. - The Trajectories Tool holds both views. Exponential curves “rarely continue for extended periods”, but if the turn takes years they “can nevertheless be highly disruptive” (2025-05-18).

2025–26: a steady middle. - His fiction takes a path that “neither panders to visions of exponential growth, or succumbs to the cynicism of hyped-up promises” (2025-11-30 postscript-letters). - Speculation about “the singularity, superintelligence and AGI is incredibly blinkered and naive”, and so is “nothing new under the sun” dismissal (2026-09-24, Q&A note). - Existential risk is unlikely but not dismissable: it would be “embarrassing if we were all wiped out by something because we didn’t have the imagination to foresee it” (2026-09-15 will-ai-really-kill-us-all).

2.6 Jaggedness, spikiness and uneven capability#

“Jagged” is a late and fairly thin node.

• The jagged model (2025-05-18). This is an S-curve with “a degree of unpredictability” laid over it. It also covers “uneven adoption of AI through society, meaning that the risks and benefits are not evenly distributed”. His example is a sentencing tool used in some court districts but not others. His prescription: “In jagged systems, traditional “set it and forget it” management doesn’t work”. What is needed is resilience and “rapid course correction”.
• Spiky surfaces and jagged edges (2025-07-27 spiky-surfaces-and-jagged-edges-moving).
• He takes Karpathy’s “Jagged Intelligence” and extends it to a “spiky” knowledge frontier: “a highly uneven frontier of knowledge”.
• Against a colleague’s smooth “knowledge closure boundary”, he builds a convoluted, fractal, n-dimensional one. On it, discovery spikes from different fields can nearly touch, and AI might act as “a catalyst or “barrier-thinner”” that helps humans “tunnel” between them. This echoes his 2021 “metaphorical quantum tunneling”.
• He is careful: this is “not … AI making new discoveries unaided here — not yet at least”. The model insists on plural ways of knowing.
• His conclusion: “Emerging AI models are challenging our understanding of knowledge and discovery”, and ignoring this risks “dangerous blindsides”.
• Related unevenness.
• The three S-curves move at different speeds (2024-12-13).
• Insiders will have “I told you so” moments while mainstream experts, who “do not grasp how disruptive the technology may turn out to be”, are blindsided (2025-01-19 wef-global-risks-2025).

Interpretation: His account of what AI is already describes a jagged profile. AI is emulation without grounded understanding, yet Evo 2’s ability to “parrot” biology “far exceeds anything humans are capable of on their own” (2025-02-23 evo-2-dna-ai). By 2026 AI is “at once deeply human and deeply alien” (2026-01-22 think-you-know-ai-think-again). He does not join these observations into a theory of uneven AI capability.

2.7 Futures thinking: plausibility, disciplined imagination, scenarios#

Plausibility is his most consistent epistemic discipline. - “what is plausible, rather than simply imaginable, is vitally important” (FFTF p.171). “humility alone isn’t enough” (p.168). - He warns against “paralysis by analysis” as much as sensationalism (2019-07-23 neuralinks-technology-is-impressive). - He calls it his own technique: let imagination run, then rein it in, “closing the shutters on hyperbolic speculation” so that ethics is “grounded in plausibility rather than hyperbole” (2024-11-17 navigating-the-ethical-dilemmas-of-brain-computer-interfaces).

Against confident prediction, in either direction. - Drastic action. Its “fatal flaw” is “the assumption that we can predict with confidence what the future will bring” (FFTF p.199). Catastrophist reasoning shares “the same conceits we see in calls for action based on technological prediction”, including “artificial certainty” (p.240). - Utopian claims. Andreessen’s “technological “foreshortening”” loses “the pain and suffering in the detail”. And “past technological successes are no guarantee of future wins” (2023-10-19 marc-andreessen-ditch-sustainability).

The Future Rising vocabulary (via 2020-10-22, 2020-12-10 and 2024-09-08). - The future as an object: a soap bubble “full of wonder and promise” that needs care. - Humans as “architects of the future” who can also “rob others of the futures they aspire to”. - Blindsides arising “in the liminal space between how we imagine the future playing out and the unknowability of what’s going to happen next” (ch. 45). - Imagination that opens pathways “even when it strays into fantasy” (ch. 20). - The need to “spot early warnings and stay clear of critical tipping points” (ch. 48).

Foresight and expert judgement. - He lists foresighting and scenario planning as neglected tools (2016-01-11). - He puts foresight on the ATT agenda (2023-09-25). - Expert surveys capture a “risk perception zeitgeist” rather than “over the horizon emergent risks” (2024-01-14 wef-global-technology-risk-trends). - Aggregated opinion tends “to regress to the mean”. That guards against speculation but discounts poorly understood risks (2025-01-19).

Scenarios as stress tests. - AI 2027 is useful for exploring “potential (if not necessarily likely) near term AI futures” (2025-04-06). - The Letters postscript weighs three scenarios: compression, superintelligence and plateau. Scenarios “allow speculative boundaries to be placed around plausible AI futures”, but “tend to focus on artificial intelligence as something that happens to society”. AI can “confound even the most prescient of futures-forecasters” (2025-11-30).

The 2026 formulation (2026-09-24, Q&A note). When “the technology changes faster than we can generate data, you’ve got to have some degree of informed speculation, and some degree of imagination”. It should be done with humility, “looking at possible futures rather than real futures”, with data to follow and “different voices”.

Who does futures thinking. Futures thinking must be “a collaborative effort — not something that should be left to an elite group of thinkers and innovators”. It is best rooted in public universities (2024-04-28). Behind this sits his question “Who gets to imagine the future and, brick by metaphorical brick, build it?”, because “the future is designed by the powerful” (2019-03-31 design-principles-for-de-marginalizing-the-future).

Interpretation: His own futures method is less formal foresight than physicist’s thought experiments joined to narrative. Examples: - Pippard’s ladder; - the petri dish and Bartlett’s beaker; - the Mandelbrot set; - “bounded infinities”, a world of infinite possibilities that may lack “the ones we need to thrive in the future”; - metaphorical quantum tunnelling; - spiky fractals.

He flags these as metaphors (“I am using this as a metaphor, no more”, 2021-04-09) and invites readers to throw them away (“the trash can of bad ideas”, 2024-08-18).

2.8 Science fiction, stories and imagination#

The core claim comes from FFTF ch. 1 and 14. Viewed critically, sci-fi films let us see “around the corner of our collective near future” (p.15). They are useful “precisely because they are not tethered to scientific accuracy” (p.288). They “cannot invent what’s yet to be discovered” and are “a poor guide to the technology itself” (p.289), but they reveal technology–society dynamics. Art is “a common point of focus” across ideological divides, “as long as that imagination is grounded in reality where it matters” (p.25).

Restatements. - Films are “pretty bad at predicting future technologies” but a catalyst for “breaking down preconceived ideas and institutionalized thinking” (2018-10-12 everything-you-wanted-to-know-about-films-from-the-future). - They are “social and educational levelers” (2019-03-07 sci-fi-movies-are-the-secret-weapon). - “transformative learning has to be felt” (2021-01-15 can-watching-sci-fi-movies-lead). - Films are “the catalyst for the journey rather than its destination”, though the book “was a failure” commercially (2023-10-08 a-guide-to-responsible-innovation). - He has taught a sci-fi film course since 2018 (2024-07-30).

Its dangers. - Dystopian fatalism, which leaves “a misplaced impression that we’re careering toward a hopelessly dystopian technological future” (FFTF p.289). - Virk’s “Sci Fi feedback loop”, in which technologists “end up ignoring reality in the belief that they can transcend it” (2024-09-18 neuralink-blindsight). - Entrepreneurs “so enamored with cool tech that they fail to spot the social messages” (2024-05-26 should-tech-entrepreneurs-be-banned-from-scifi). - Transcendence serves as his object lesson in exponential make-believe (2018-11-15).

Stories. Stories are “the pivot point between being able to imagine the future and beginning to build it” (Future Rising ch. 29, quoted in 2024-01-21 how-can-stories-unlock-pathways-to). They open minds where “Preach to someone about the future, and most people will shut down”. The darker mirror: who will create “the stories that determine our beliefs” (2024-09-22)? That question links this thread to his manipulation thread.

Fiction as his own method. - Letters from the Department of Intellectual Craft (2025). - “Soul Update” (2026-02-11). - A Claude Code-assisted study of 169 AI films (2026-05-15 ai-movies-may-be-less-dystopian, his prose only). He reports that 32% are dystopian, that the dystopian share is declining while “protopia” films rise, and he stresses cases with “no bright line between artificial intelligence and a particular future state”.

Imagination in education. In the FRANKx lecture, the arts, humanities and “alternate ways of knowing” help us escape a “bounded infinity” of conventional thinking. He designed Future Rising itself as a “quantum tunneling enabler” (2021-04-09).

2.9 Inevitability, steering and whose future#

Inevitability hardens over the years. - 2018. Renouncing technology “from a position of privilege” denies others choices (FFTF p.288). Neural interfaces: “Not that I think this should be taken as an excuse not to build” (p.146). - 2024. “we cannot put the genie back in the bottle” (2024-04-14). “we are already irreversibly integrating AI into every aspect of our lives” (2024-07-21 artificial-intelligence-dune-villeneuve). - 2025. “Technology is not deterministic — it doesn’t just happen to us” (2025-03-30). “The AI genie is out of the bottle”, but it can be channelled “much as a flood can’t be halted, but it can be directed” (2025-08-31 holding-on-to-our-humanity-age-of-ai). - 2026. “the boat has already left the harbor here” (2026-05-21). “We can’t run away from it. We can’t stop it. We can’t pause it.” (2026-09-24). In the same lecture he criticises the race logic of “if we don’t go fast, somebody else will”.

Steering is always tied to whose future. - We risk “handing our futures to innovators whose vision exceeds their understanding” (2021-09-07 should-we-be-worried-about-elon-musks-tesla-bot). - Foreshortening hides “who decides who will suffer and who will thrive” (2023-10-19). - Public indifference hands power to “those that do care” (FFTF p.286). - We must ensure “our technological reach doesn’t exceed our collective grasp” (2019-04-26 avengers-endgame).

3. How the thread developed#

Constants. - Complex systems are unpredictable in detail but bounded. - Irreversibility rises as connectedness grows. - Plausibility is his discipline, applied to hype and to doom alike. - He never drops the physical-limits caveat on exponentials. - Sci-fi is a lens, not a forecast. - He favours steering over stopping. - He keeps returning to who imagines the future, and who pays for it.

Shifts. 1. From convergence to AI. Convergence organised his thinking in 2015–21. AI later becomes the accelerant within it. 2. Exponentials. A fallacy to debunk (2018) becomes also a blind spot to plan around (2025). This is movement toward planning “on the off chance”, not a conversion. 3. Tipping points. To be avoided (2023); possibly to be embraced (2024). 4. AI’s status. One strand among many (to 2021); conditionally a different category (2025); beyond analogy (2026). 5. Timescales. Consequences outpace fixes (2021); disruption in months (2023); responsible innovation possibly “futile” (2025). 6. Method. A 2018 Occam’s Razor ranking gives way to tail scenarios and “informed speculation”. 7. Institutions. Confidence moves from government to universities, then to disappointment in universities.

4. Connections to his other threads#

• Risk innovation and threat to value.
• The threat/opportunity transitions model is built on threat-to-value (2024-08-25).
• The Trajectories Tool defines “effect” as value or care (2025-05-18).
• Orphan risks were first justified by complex-systems reasoning (2020-11-05).
• “Novel theories, models, and approaches to risk” is an ATT research domain.
• Complexity is his core reason that conventional risk tools “belong to a different world” (FFTF p.23).
• Responsible and permissionless innovation.
• Irreversibility and complexity are his main arguments against permissionless innovation (“a world made of kindling”; the reversibility test).
• The timescale mismatch is his main doubt about responsible innovation under acceleration.
• Manipulation, cognition and formation.
• Joy’s self-replication becomes self-replicating ideas spread by language-fluent AI (2023-04-26).
• Stochastic agency is an emergence argument.
• The companion bot is his paradigm “chaotic” case.
• The tipping point he fears most changes “who we are”.
• Being human.
• Technology as constitutive (the oxygen analogy).
• The three domains: what we do, where we live, who we are.
• Future Rising: envisioning and changing the future is “part and parcel of what it means to be human” (ch. 53).
• Who decides and justice. Whose future, stolen futures, foreshortening, collaborative futures thinking, and the unanswered question in the Embrace quadrant of who thrives.
• Existential risk. The plausibility filter, the thermodynamic doubt and catastrophe reframed as loss of value all come from this thread.
• Nanotechnology biography. His scepticism of exponentials was formed in the gray-goo and nanobot debates (the Bostrom dinner, the ITS bombing), and his wariness of revolution narratives in “brand nanotechnology” (FFTF p.214–218).
• Education.
• A proposed school of ATT.
• “How to build positive futures” as a durable learning area, which he says means teaching “how to navigate advanced technology transitions” (2024-03-10 harari-teaching-future).
• The sci-fi film course.
• Educating “base coders” and “metaphorical quantum tunnelers” (2021-04-09).

5. Tensions, ambiguities and gaps#

1. Past lessons against present novelty. This is the key tension for anyone using his work as a lens. - He complains that each wave “re-invent[s] the wheel” (2023-04-12) and wants historic failure modes studied (2023-09-25). - Yet he calls the present “night and day different” (2024-08-11), says AI “defies analogy” (2026-01-22), and says past frameworks yield “categorical errors” (2026-09-24). - Interpretation: His implicit resolution is that past transitions supply structural and process lessons, not templates. He keeps using analogies (Pippard, chaos, the beaker, gray goo, viruses) while denying that any one captures AI. He never states this resolution.

2. Plausibility against unpredictability. - Occam’s Razor discounts scenarios built on stacked assumptions (FFTF p.281). - Complexity says small, unforeseen events trigger discontinuities, and blindsides live in what we fail to imagine. - By 2025 he plans for low-probability, fast-moving tails and warns we will deny them “until it was too late”. - He offers criteria (physical law, humility, data to follow, many voices) but no operational test separating a disciplined edge case from “make-believe treated as plausible reality”. He does not revisit his 2018 Razor.

3. Plateau scepticism against urgency. He forecasts “no AGI emerging in the near future” (2023), doubts scaling, and thinks an S-curve ceiling most likely. Yet he calls for urgent, billion-dollar preparation (2025-01-07). The risk logic reconciles the two: a small chance of a large effect justifies preparing. But it lets him hold both, and leaves his central expectation for AI underdetermined.

4. Inevitability against choice. “We can’t pause it” (2026) sits beside his 2024 call to consider “pausing — or even rethinking” companion chatbots (2024-10-27), beside the legitimate “Avoid” quadrant, and beside “Technology is not deterministic”. Interpretation: He seems to treat inevitability as a property of the overall trajectory and choice as a property of particular designs and uses. He never says so.

5. Embracing tipping points against justice. The Embrace quadrant asks us to cross irreversible transitions deliberately. He admits he does not know what that means for “who thrives and who does not”. That fits badly with his critique of foreshortening, and with his view that no one has the right to act unilaterally on futures that affect everyone (FFTF p.249). He has not developed the quadrant since 2024.

6. Science fiction as lens and as hazard. Sci-fi widens imagination, but it also: - breeds fatalism; - feeds the “Sci Fi feedback loop”; - supplies “make-believe” that can do harm.

His 2026 corpus study challenges the dystopia trope, but by his own account it does not show how films shape actual futures. Outside his classroom, it is unclear how the “right way” of viewing is secured.

7. “Jagged” is underdeveloped. He uses it for uneven adoption and a spiky knowledge frontier, not for AI’s uneven capability profile, although his account of AI’s nature invites that use. His language about AI also jumps from “stochastic parrot” (2025-02-23) to “Rather, they are different” (2026-01-22) with no explicit bridge.

8. Convergence recedes; tools stay heuristic. - His strongest early claim was that technologies must not be assessed in isolation (2015). His post-2023 analysis is largely AI-centred. - The ATT tools remain quadrants and curves. - His institutional proposals (a school, a cross-agency initiative, billions in philanthropy) are ambitious, but their mechanisms are thin. By 2026 he says universities have been “followers and users of the technology” (2026-08-30). - He names foresight, scenario planning and technology assessment as needed, but seldom uses them himself. He engages little, by name, with futures-studies literature.

6. Most important sources for this thread#

1. FFTF (2018), ch. 2 (pp. 39–43), 8 (pp. 163–178), 9 (pp. 179–206) and 13 (pp. 279–286). Convergence and base code; bounded chaos; irreversibility; the exponential and singularity critique; make-believe harms; Occam’s Razor; sci-fi as a way of seeing.
2. 2015-01-30 responsible-development-of-new-technologies-critical-in-complex-connected-world. Converging technologies as a fragile system; against assessing technologies in isolation.
3. 2018-11-15 even-bad-sci-fi-movies-can-teach-us-something-about-emerging-technologies. Exponential extrapolation and speculative fear causing harm.
4. 2021-02-25 how-our-mastery-of-biological-physical-and-cyber-base-code. Base code as the pivot of history; “bricking”; social base code.
5. 2021-04-09 bounded-infinities-quantum-tunneling-and-the-future-of-education. The solution problem; the inversion of timescales; bounded infinities.
6. 2023-04-12 navigating-advanced-technology-transitions, with 2023-09-25 building-a-better-futures-tough. ATT named, defined and turned into a research agenda.
7. 2023-05-04 tipping-points-and-broken-symmetries. Pippard’s ladder; the past as a poor predictor.
8. 2024-03-31 we-have-a-technology-problem-and. Technology as constitutive; pilots.
9. 2024-08-18 four-ways-of-thinking-about-advanced-technology-transitions, with 2024-08-25 advanced-technology-transitions-model. The four quadrants; the threat/opportunity model; the inevitability assumption.
10. 2024-09-08 a-journey-from-the-past-to-the-edge-of-tomorrow. His Future Rising statements on complexity, acceleration, the singularity, blindsides and stories.
11. 2025-03-30 reimagining-education-in-an-age-of-ai. What we do / who we are; three trajectories; guide and steer.
12. 2025-04-06 responsible-innovation-and-ai-acceleration (his framing essay only). Exponential blindness; the timescale mismatch; scenarios as stress tests.
13. 2025-05-18 exploring-ai-through-cause-and-effect. Six cause–effect models, including jagged and chaotic.
14. 2025-07-27 spiky-surfaces-and-jagged-edges-moving. The spiky knowledge frontier; AI as barrier-thinner.
15. 2026-01-22 think-you-know-ai-think-again, with 2026-09-24 being-an-academic-in-an-age-of-ai (Q&A note). AI beyond analogy; informed speculation; “can’t pause”.

Runners-up: - 2023-10-19 marc-andreessen-ditch-sustainability - 2023-11-29 the-year-that-generative-ai-changed-the-world - 2024-04-28 beyond-the-future-of-humanity-institute - 2025-11-30 postscript-letters-from-the-department-of-intellectual-craft - 2026-05-15 ai-movies-may-be-less-dystopian-than-we-think