Late Lessons, Jensen Huang and AI
Markdown version

B16 perspective notes: 2024-06-25 to 2024-08-18 (15 posts)#

These notes read the batch for how Maynard thinks, not for the concepts he names. All fifteen posts were read in full. There are no Modem Futura note posts in this batch. Quotes are exact, including his typos (“I’d of laughed”, “raised my wife’s craft supplies”, “vey much”, “plans” for plants).

Evidence rules applied

Context. It is summer 2024. He writes from Dalian at the WEF “Summer Davos”, where he has contributed to the Top Ten Emerging Technologies list since 2012. He is preparing, and then gives, a keynote in Bali to the IEEE International Symposium on Consumer Technology, titled “A New Science of Navigating Advanced Technology Transitions”. Four posts deal with Elon Musk’s ventures. “Advanced technology transitions” becomes the name for his agenda across the batch, and “navigating” is its verb.

1. How he thinks here#

He opens from where he is and what he is doing#

Most posts come straight out of his working life. The writing reads as a notebook kept in the open, not as commentary from outside:

The big questions are reached from these particular starting points, not announced as theses.

Building to think: the Lego ladder#

The clearest instance of his method in the batch is 2024-08-18 four-ways-of-thinking-about-advanced-technology-transitions. It runs as a sequence of moves:

  1. Owning a long fascination. “I have a thing about “Pippard’s ladder.”” The ideas are ones “I’ve been playing with for some years”.
  2. Starting from a remembered demonstration. He goes back to the moment as a PhD student when a physicist’s public demonstration showed that “our vision of the future was utterly wrong” (quoting his Future Rising ch. 48).
  3. Asking whether the metaphor can do more. “I was interested in whether it could be extended to thinking about different ways of approaching an uncertain future.”
  4. Building it with his own hands. “I visited our local Lego store, raised my wife’s craft supplies, and created my own rather crude (but nevertheless serviceable) ladder.”
  5. Letting the object generate the idea. “Experimenting with the ladder while thinking through the concept of advanced technology transitions” is what produces the four-quadrant framework. The framework comes out of play; he did not design it first and illustrate it afterwards.
  6. Demonstrating each quadrant physically on video. Stabilising the ladder “using clothes pegs!” stands for adaptation. Extending the ladder stands for pushing boundaries back.
  7. Holding the result lightly. “It may be so deeply flawed that it should be resigned to the trash can of bad ideas.”

The physical, playful, provisional quality is the method, not decoration. He reports the exclamation mark and the craft-box improvisation with evident pleasure.

Simple models as thinking devices, then made dynamic#

He draws two quadrant diagrams in the batch (Choice Engines, Pippard’s ladder), and both are introduced self-deprecatingly: “quickly sketched out”, “admittedly rather simplistic”, “admittedly simplistic”. The move that matters is what he does next. He turns the static 2x2 into a field with motion in it:

This is a physicist’s habit: build a toy model, then ask which way things will flow. He borrows the language of physics (gradient, forcing function, degrees of freedom, nonlinear dynamics) to reason about political economy and society.

“What if?” as the turning point#

The key turns in the ladder post are questions:

He admits the second takes him somewhere he cannot yet see: “to be honest I’m not sure what embracing a disruptive technology-driven tipping point might look like”. He follows the question anyway, because following it is how the thinking moves.

Plausibility tests from first principles#

He checks grand claims against how physical and biological systems actually behave:

Reframing where the real problem lies#

Again and again he moves the problem somewhere others are not looking:

Small imagined scenes#

To test the robot future he uses a domestic, embodied scene rather than statistics. You go to the kitchen at night and glimpse “an unexpected human-like figure”, or “you wake up to see your humanoid robot seemingly watching you from the end of the bed” (2024-08-07). The response is “driven by biology and instinct, not rational thought”. This is the same move he makes with films: a concrete, felt scenario that exposes what an abstract analysis would miss.

Films and stories as ways in#

Holding tensions without resolving them early#

Curiosity and delight drive the attention#

Serendipity is named, and it is also how he picks topics: a side-meander in an eight-hour podcast becomes a section of the post.

2. What matters to him#

Human agency and self-determination#

This is the thread that runs most consistently through the batch:

What makes us “us”#

The question under the Dune review is what AI might diminish in “who we are”. His answer is resilient but not complacent: “humanity is sufficiently adaptable and resilient to hold onto what makes us “us””, yet “this does not mean that the question of how AI could diminish who we are” can be set aside (2024-07-21). Musk’s enhancement vision “should absolutely be part of a much wider conversation around where we collectively want to go as a species” (2024-08-04).

Inclusion and access, “for as many people as possible”#

Learning over education#

He wants transformation, not throughput. The efficiency model of teaching “places quantity before quality” and puts “outputs (such as assignments and grades) ahead of outcomes (such as transformed lives)” (2024-07-30). Education and learning are central to navigating transitions: the movies class, the school concept, and his interest in Ad Astra as a “post-AI” experiment.

Humility and honesty in knowledge#

He praises Seth for writing with “humility”, because Seth “readily admits that his ideas may be wrong, and invites people to test and build on his thinking”. That humility is “essential to good science and responsible progress” (2024-06-30). He applies the same standard to himself (“I may be wrong”, 2024-08-07) and to advocacy data (2024-07-28).

What frustrates him#

What delights him#

3. Risk as a way of thinking#

Risk as a threat to value: whose value, created or taken#

2024-07-13 ai-choice-engines-sunstein states the frame outright. The question is the “could” of AI Choice Engines against the “should” of technologies that “risk threatening value in unexpected ways”. His whole analytic device is built on value, with axes for value to the individual and value to the deploying “agent”, not on hazard. Risk here is the question of where value flows and who loses it. It is structural, not malicious: “This may not be intentional or even malicious.” The concept does work: it shows that even the apparently benign case, a government nudging diets, slides toward manipulation, because an agency’s own value includes “power and influence”.

A risk landscape with slopes#

The Choice Engines diagrams amount to a risk landscape in all but name. Positions are quadrants, forces are “forcing functions”, and there is a slope, the “economic gradient”. The response he calls for is navigational: “new thinking on how to resist the economic gradient toward manipulation, and how to push AI Choice Engines up toward the empowerment quadrant or the transformation quadrant”. He describes the whole challenge as “navigating the emerging capabilities of AI to influence personal decisions”. The picture is of steering across terrain, not setting a control limit.

Novel technology demands a new mindset: the explicit claim#

Beyond avoidance: a change of mind about his own metaphor#

The Future Rising excerpt he reprints ends with the earlier lesson of Pippard’s ladder: “unless we learn how to spot early warnings and stay clear of critical tipping points, we run the risk of, quite literally, crashing our future.” In 2024 he says this avoidance reading “is vey much in line with my original thinking”, and then goes past it:

He does not rank precaution below the rest. He places it as one of four postures and argues that thinking only in the lower two leaves possibilities unexplored. The model’s stated function is “to stimulate new thinking”.

Perceived safety and the sacrosanct as the real terrain#

In 2024-08-07 are-humanoid-robots-really-the-future:

He does not treat perception as an error to be corrected. He treats it as the ground any innovator must actually cross. He also predicts that the enterprise’s value is most threatened by social factors its builders are not attending to. His advice is to “build robust teams that focus on overcoming the societal challenges they face”. This is the logic of risk innovation and of orphan risks, though neither term appears.

Quantitative grounding stays in view alongside this: “millions of hours of simulation and training”, and more “degrees of freedom” than autonomous vehicles.

Overlooked risks#

He spots risks that fall outside the usual categories:

Humility against false precision#

Inevitability plus steerability#

He assumes innovation will not stop. “technological innovation and advances are inevitable given our human nature” (2024-07-13). “we cannot simply turn off technology innovation, but instead need to find ways to manage and channel the inevitability of technological change”, which he anticipates “some will disagree with” and is “comfortable with” (2024-08-18). But direction is open: we can “steer it toward what we want” (2024-07-21). This pairing of a current that cannot be stopped with a course that can be chosen is why “navigating” fits better than either “managing” or “stopping”.

How these concepts function. In this batch they work as mental models, not procedures:

Their job is to widen the set of options people can see. The only near-operational prescriptions are brief: guardrails and checks and balances, and societal-challenge teams.

4. Scholarship and public writing#

Ideas move openly between formats and years#

Substack is where his scholarship is drafted, tested, brought back and reconnected.

Correcting in public#

Evidence and expertise#

Transdisciplinarity as a structural commitment#

Accessibility#

Demonstration as communication#

He learned about tipping points from Pippard’s public lecture with a physical model. He now does the same thing for an audience of consumer-technology engineers, with a Lego ladder on video (2024-08-18). A playful, physical, public way of teaching is carried down a lineage.

5. His role as he sees it#

Inside the rooms, as a participant rather than a commentator#

Pilot and guide#

The school’s self-description reads as his own account of the public scholar’s job. It would “take seriously the idea of serving as a guide or “pilot” for individuals, communities and organizations as they face technology transitions that demand radically new ways of thinking” (2024-08-11). Its faculty would be “creative, innovative, entrepreneurial” and aim to be “prominent voices guiding decision-making”.

Conversation starter, not oracle#

He repeatedly frames his contributions as openings:

For his students#

He attends to Musk because it is “deeply relevant to my work and to my students” (2024-08-04). Students recur as the people the thinking is for: the movies class, the school, Ad Astra, and the transdisciplinary aside.

With people he disagrees with: fair, generous, not captured#

The Musk posts show this most clearly:

Toward Neuralink he credits the establishment’s concerns (“Some of this is most likely justified”, and he links his own earlier critique). He also teases the establishment with the aside that, “most infuriatingly”, Neuralink brings a disruptive mindset to medicine. On Ad Astra he weighs “baggage” against “a growing need for innovative approaches to learning” that holds “irrespective of who’s backing” it.

One tension is worth recording. In 2024-06-30 he criticises the “hubristic proclamations” of AI proponents, and weeks later he enjoys Musk’s “flights of speculation”. The posts suggest how the two fit together. He values bold speculation as a vision to engage with and prepare for. What he rejects is certainty without humility, and the “go fast and break things” attitude when it is backed by power and “very little governance”. He wants an “informed counterbalance”, not a ban on dreaming.

What he refuses to do#

Changes of mind, shown in the text#

6. What is distinctive#

  1. Thinking by making, in public, with play. A tenured physicist-turned-risk scholar raids the craft box, builds a Lego ladder, pegs it with clothes pegs, films it, and presents it to an IEEE keynote audience as a serious framework he is ready to throw away. Few commentators on AI generate ideas this way or show it.
  2. Physics structure applied to social dynamics. Gradients, forcing functions, degrees of freedom, thermodynamic equilibrium, nonlinear tipping points: he uses them as structural analogies to see which way things will flow. The Choice Engines “economic gradient” joins political economy to physical intuition and yields a non-obvious conclusion: benign government nudging slides toward manipulation too.
  3. Mindset made an explicit variable. His transitions framework puts mindset on an axis, which makes his claim that novel technologies need a different way of thinking visible and open to argument. He places precaution as one legitimate posture among four, not the default and not a straw man.
  4. Perception and the sacrosanct treated as the actual terrain. He does not dismiss public unease as irrational. He predicts it will decide outcomes, as in the robot-in-the-kitchen scene, and advises industry to invest in the social side. Almost no one else in the humanoid-robot discussion was saying this.
  5. Fiction as a governance instrument. A review of Dune: Part Two in a law journal that reads AI’s absence on screen as a prompt for thinking about AI governance and human identity.
  6. Serious, generous, critical engagement with figures others either worship or dismiss. He reads Musk as a force shaping the future that must be understood, and says openly that he enjoys him.
  7. Agency as the value under the risk analysis. Choice Engines, UBI, robots and Dune all come back to whether people remain the authors and beneficiaries of their own lives. His risk thinking is built around that value, not around hazard counts.
  8. Education as the main lever for transitions. Where others reach for regulation, he reaches for new kinds of schools, courses and learning spaces, with films watched together as a core method.
  9. Evidence humility used in both directions. He is sceptical of advocacy numbers he agrees with, respects qualitative stories, and asks whether measured indicators capture what matters.

7. The posts in this batch that best show how he thinks#

  1. 2024-08-18 four-ways-of-thinking-about-advanced-technology-transitions. His method in one post: a remembered physics demonstration, play with Lego, a model that emerges from the play, “what if” turns, a framework with mindset as an axis, a change of mind about his own metaphor, precaution placed fairly, and a model offered as disposable.
  2. 2024-07-13 ai-choice-engines-sunstein. He extends another scholar’s argument. He is explicit about risk as a threat to value, turns a 2x2 into a landscape with gradients, finds the non-obvious result that governments slide too, and grounds the whole thing in agency and anti-paternalism.
  3. 2024-08-07 are-humanoid-robots-really-the-future. A reframing from technical to societal. Safety is perception and the sacrosanct. An embodied scenario, complex against complicated, a self-reported change of mind, and advice to industry.
  4. 2024-08-04 7-key-takeaways-from-elon-musk-and-lex-fridman. How he treats a polarising figure: curiosity, plausibility checks, delight in speculation, criticism by name, an “informed counterbalance”, readers warned that he enjoyed it, and accessibility work (clips and timestamps).
  5. 2024-07-21 artificial-intelligence-dune-villeneuve. Film as a mirror. The scrappy-post-to-journal cycle shown openly. The refusal of polemics and fear, with resilience affirmed and the question kept open. Scholarship that sits in a law journal and on Substack alike.
  6. 2024-08-11 school-of-advanced-technology-transitions. His own statement of the premise that novel technologies are “not navigable through conventional thinking”, and of the public scholar as “pilot”. It also shows transdisciplinarity built into structure, and conversation-starting as his mode.

Runner-up: 2024-07-28 massive-new-study-reveals-new-insights-into-ubi is the best single example in the batch of how he handles evidence (noisy signals, measured against affected, advocacy bias, stories as windows) and of agency as the value he returns to.