Future of Being Human an Arizona State University initiative

Learning by not trying to learn: purposeless play as a professional skill

Last updated 2026-08-07 · Markdown version

Learning by not trying to learn is the central thesis of Andrew Maynard’s 2026–2027 sabbatical, stated publicly on 2026-08-02: purposeless play, joy, delight, and serendipity are critical professional skills for thriving amid transformative AI and other transformative technologies — and environments deliberately free of learning objectives can produce learning that instrumented education cannot. Its working demonstration is Hyperbubble, a free browser game devised, designed, and coded by Anthropic’s Claude Fable 5 in an extended collaboration directed by Andrew, built to a brief whose hard constraint was that it “must not feel like learning, but should implicitly lead to new learning.” This is active work — an evolving research program, not a finished doctrine — and this page tracks it as it develops.

The argument

The claim arrived as a confession about what sabbaticals do to the unwary: given time to think “unhindered by the pressure to produce,” you may find yourself pursuing the idea that play without purpose, embracing what sparks joy, and reveling in the small delights of unexpected discovery are “all critical skills for thriving in an age of AI” (2026-08-02). Two claims are folded together. The first is a skills claim: play, joy, delight, and openness to serendipity are not the opposite of professional competence but part of it. The second is an environments claim: settings deliberately freed of learning objectives, measured outcomes, and assessment can produce learning that instrumented education cannot. Andrew is explicit that neither is novel — “For people who study and embrace such ideas, of course, this isn’t new” — and that he has inhabited this space for the past decade or so. What the sabbatical adds is seriousness: treating these ideas as a research program about thriving professionally as transformative technologies rewrite “the rules of how we do pretty much everything.”

The thesis gets its bite from a diagnosis. Professional culture, he argues, devalues play: how we teach, how we develop career-enabling skills, how we evaluate ourselves and others, and how we behave in professional environments all tend “to devalue and discount the importance of joy, delight, and play, and to treat them as trivial, immature, and not appropriate for serious people doing serious jobs” (“If you doubt me, just take a quick look at your LinkedIn feed”). He hedges the diagnosis himself, and the hedge is part of the argument: “I suspect many people will push back” — lip service abounds — but apart from a few select professions and organizations, “actions rarely indicate that ideas like joy, delight, and play, are treated with respect” (2026-08-02, footnote 2). The claim is about revealed behavior, not stated values.

Hyperbubble is the working demonstration. A one-button browser game — a soap bubble (a reference to his book Future Rising) riding an undulating hand-drawn future, adopting orphan risks, managing emerging tech risks, not feeding moral panics — built in the Claude Fable 5 collaboration described above (Tools and experiments carries the full provenance). What matters here is the brief: Fable’s published self-audit records, among Andrew’s constraints, the one it calls “the hard one” — the game “must not feel like learning, but should implicitly lead to new learning” (2026-07-10). By August 2026 it had reached version 5 through more than 90 iterations, pursued “partly out of the simple joy and delight of doing this” and partly to explore the thesis itself (2026-08-02).

The mechanism claim is precise: absence of learning objectives is not absence of design. The game was designed “with great care … and with substantial input from AI” to encourage learning through serendipity, joy, and delight — yet nothing in it depends on programmed experiences, learning objectives, or documented outcomes. It is “a free space for play and exploration. A playground,” and what a player takes away “is uniquely theirs — and not determined by a set of learning outcomes” (2026-08-02). The bet: remove the demand that learning occur, and conditions form in which it occurs naturally — about how he thinks and sees the world, about navigating advanced technology transitions, about flourishing in a technologically complex future. He reports the learning ran through two channels, one unexpected: developing the game with Fable changed his thinking, “But it’s also been influenced by actually playing the game — which is something I wasn’t expecting.”

Then the essay does something easy to misread as weakness: it undercuts itself. “Of course, I’m just messing around here, and am probably over-stretching the significance of the game … But that, of course, is the point.” The self-undercut is load-bearing. A claim that purposeless play produces learning cannot prove itself with pre-registered outcomes without becoming the thing it criticizes; it can only document instances, preserve its own uncertainty, and invite the reader into one. The register — whimsy carried on a career of risk and transitions scholarship — is the method, not a lapse from it.

The stance predates the sabbatical. In AI and the Art of Being Human (with Jeff Abbott, 2025), “Purposeful Purposelessness” is a named design feature of gathering spaces — “You don’t attend to network or optimize — you come to be human together. And from that, meaning emerges” (p. 167) — and the book’s STARS practices hold that “the radical act is presence without progress” (p. 127). His machine-readable index files the lineage as method: “Play and popular culture as serious methods” (andrewmaynard.net/llms.txt, as of its 2026-08-04 update). Where Playgrounds, not playpens makes the education-design case for students, this thesis aims at professional life — and he expects to write more as the sabbatical proceeds.

Lineage

In his own words

“… like pursuing the idea that play without purpose, or embracing what sparks joy, or even reveling in the small delights of unexpected discoveries, are all critical skills for thriving in an age of AI. And that, sometimes, the best way to learn and grow when transformative technologies are rewriting the rules of how we do pretty much everything, is to not try to learn.”

— What we can learn with AI by NOT trying to learn (AI-readable mirror), 2026-08-02.

“Because the thing I keep coming back to here — and what is increasingly part of my thinking as I continue with my sabbatical — is that the magic of Hyperbubble is that there are no learning expectations. The game doesn’t stand or fall on programmed experiences, on learning objectives, or measured and documented outcomes. It’s a free space for play and exploration. A playground. Somewhere where you can be whatever you want to be, and play however the mood takes you.

It’s an environment that’s been designed — with great care I have to say, and with substantial input from AI — to encourage learning through serendipity along with joy and delight, and to contribute to the formation of a mindset that is attuned to thriving in a technologically complex world. But what a player takes away from it is uniquely theirs — and not determined by a set of learning outcomes.”

— What we can learn with AI by NOT trying to learn (AI-readable mirror), 2026-08-02.

“Of course, I’m just messing around here, and am probably over-stretching the significance of the game and the process that led to it.

But that, of course, is the point. Especially while I’m on sabbatical!”

— What we can learn with AI by NOT trying to learn (AI-readable mirror), 2026-08-02 — the essay’s closing lines.

Engagement and reception

The thesis essay posted on 2026-08-02 — four days before this page — and Hyperbubble first launched 2026-07-10 (per Andrew, 2026-08-07); our reception record documents no independent engagement with either yet, and we prefer to say so plainly. That is an absence of record, not a verdict on the idea: conceptual work of this kind typically accrues visible reception slowly, and we don’t typically solicit it (Philosophy). What is checkable now is process, all self-published: the Hyperbubble posts document the build publicly, including Fable’s self-audit of its own limitations (2026-07-10), and the game is live with a global leaderboard. One earlier lineage item carries its own dry reception datum: by Andrew’s account, the 2021 shipment of books to the Tesla Bot team received no acknowledgment at all (2023-08-28).

Where to go deeper

Full essays (oldest first, each with canonical and AI-readable links):

The game and the book

Related ideas-ring pages

Related corpus pages