Table of Contents
The silence
I owe you an explanation.
The last post on this blog was published in May 2022. Nearly four years ago. In internet terms, that’s roughly the Cretaceous period. In AI terms, it might as well be pre-Big Bang.
I could give you the usual excuses — busy teaching schedule, life getting in the way, the blog perpetually sitting at number three on the to-do list behind “mark Physics coursework drafts” and “get back in the cricket nets”. All of those things are true. But if I’m being honest, the deeper reason is that somewhere along the way, the Rambo bandana got a bit loose, and I lost sight of the forest for the trees.
Then, a few weeks ago, I read an article that had, by some counts, reached 83 million people, and something clicked back into place.
Something big is happening
Matt Shumer’s viral LinkedIn post — Something Big Is Happening — is the kind of thing that either sounds like the ravings of a tech evangelist who’s been drinking too much Silicon Valley Kool-Aid, or a genuine wake-up call, depending on where you’re standing. He opens with a comparison to February 2020 — the month most of us first heard the word “coronavirus” and thought it sounded overblown. We all know how that played out.
His central thesis: AI hasn’t just gotten smarter — it’s crossed into genuinely useful territory for real work. Not as a search engine replacement. Not as an autocomplete on steroids. Something closer to a capable junior colleague who can handle serious tasks if you give them the right brief. Most people, Shumer argues, are still using it the way they used Google in 2004 — typing a question and reading the result — when they should be pushing it into the heart of their actual work.
His practical advice is simpler than you might expect: sign up for the paid version of Claude or ChatGPT. Don’t dip your toe in. Dive in. Start with the thing you spend the most time on and see what happens.
I read this sitting in Colombo, between marking a stack of MYP mathematics quizzes and wondering why my Claude-generated physics test document had the equations rendering in the wrong font. Again. I signed up for Claude Pro that same week — partly out of genuine curiosity, and partly to rule out the possibility that I was simply missing out by not using the paid version.
And honestly? My first thought was: Matt Shumer has clearly never tried to get an AI to produce a properly formatted IB Physics test with labelled diagrams.
But my second thought was: he’s not completely wrong either.
The fork, according to Diamandis
A few weeks later, Peter Diamandis published a Substack post called Humanity Is About to “Fork” — and it was essentially the second push that motivated me to finally sit down and write this article.
Diamandis argues that humanity is approaching not one but five major forks: Creators vs. Consumers, Longevity, Brain-Computer Interfaces, Earth vs. the Stars, and Digital Consciousness. While I love nothing more than speculating about those relatively far-future possibilities — and probably more than most people do — I’m also a pragmatist, and a firm believer in taking actions within your sphere of influence to optimise your own present and near-future (see my post on optimising the happiness curve for more thoughts on this). And so it was this passage about Creators vs. Consumers — “Fork 1” — that slapped me in the face like a raw fish:
“AI has handed every human being on the planet an extraordinary set of tools: the ability to build software, design products, generate content, start companies, and pursue ambitions that previously required teams of specialists and millions in capital. Some people will pick those tools up and build. Others will watch.”
And as if to underline the point, Diamandis published another piece this week — on whether anyone can become an entrepreneur in the age of AI. His closing line stopped me in my tracks again: “The tools are free. The information is free. The only question left is: What are you going to build?” Uncanny — because six years ago, standing next to an eBike in Myanmar, I wrote something almost identical: “You have all the tools you need to learn whatever you want to learn and improve your individual situation at your fingertips. The question is, what are you going to do with them?”
Apparently great minds think alike — even six years apart.
The honest audit
Here’s the thing about Fork 1 that the hype merchants tend to gloss over: picking up the tools doesn’t automatically mean the tools work the way you hoped.
I’ve been using Claude Pro seriously — not as a search engine, but pushed into my actual work. Generating exam-style questions, performing detailed analyses of student lab reports against rubrics to make sure my marking calibration is spot-on, and vibe coding interactive HTML simulations to help students understand concepts like entropy and microstates.
The verdict after several months? Complicated.
Where Claude genuinely moves the needle: deepening my own understanding of the hairier corners of IB Physics. Special relativity, thermodynamics, quantum mechanics. Having a thinking partner that can work through a Lorentz transformation problem, or help me build a rigorous derivation without reconstructing everything from scratch — that’s genuinely valuable. The quality of my teaching resources has improved measurably.
Where Claude hits a wall: producing the actual finished document. Equations rendering incorrectly. Diagrams pointing the wrong way. Hours spent iterating on font sizes and arrow directions when I should be thinking about physics. You know that opening scene in Rambo: First Blood Part II where Rambo is breaking rocks in a prison quarry with a sledgehammer? That’s what generating a properly formatted two-page quiz can sometimes feel like — except at least Rambo eventually got offered a mission.
The time savings that Shumer promises? Not yet. Not for my workflow. The tools are at something like 70-75% of the way to genuinely useful for document production — and that last 25% is eating all the time savings from the first 75%.
But here’s the reframe I’ve landed on, borrowed from Diamandis himself: he doesn’t say “I’ve never worked less in my life.” He says “I’ve never worked harder in my life, and I’ve never had more fun doing it.” That’s a different promise entirely. And that, if I’m honest, is closer to my experience — at least most of the time, when I’m not fighting with formatting, and I’m bringing ideas to fruition that simply wouldn’t have been possible before.
The 80/20 rule revisited
Regular readers of this blog (both of you) might remember my 2020 post on the art of bodging and the 80/20 rule. The sigmoid curve of effort versus results. The argument that it’s rarely worth pushing from 80% to 100% when you could be moving something else from 0% to 80%.
I’ve been violating that principle badly with Claude.
The content generation — the physics questions, the conceptual explanations, the derivations — that’s the steep part of the curve, where Claude delivers enormous value per unit of effort. The document production — the formatting, the diagrams, the rendering — that’s the flat part at the top, where I find myself shaking my fist at the screen and demanding that an entity with no emotions fix its arrows. As Seneca noted in De Ira, that’s about as useful as kicking a mule.
The smarter play, which should have been obvious from my own prior writing, is to let Claude do what it does well and stop expecting it to do what it doesn’t yet do well. Wait for the tools to mature — probably another six to twelve months, if Shumer and Diamandis are to be believed — and in the meantime, extract the leverage from the part of the curve that’s actually steep.
Which side of the fork?
So where does this leave the STEM Warrior in 2026?
I’m on the creator side of Fork 1. Unambiguously. Building interactive simulations. Developing original teaching resources. Running my DP Physics course around a framework I call Physics of the Future — which, now that I think about it, sounds like a pretty good name for a podcast. Thinking seriously about what AI-enhanced STEM education actually looks like from inside a classroom — not from a conference stage, not from a LinkedIn post, but from the coalface of a Grade 11 Physics class in Colombo, where some students are hungry to dive into quantum mechanics, and others are mainly wondering whether AI will one day write their lab reports for them.
Moving forward, I want to carve out more time for the creator-side activities — the simulations, the resources, the writing, and yes, possibly a podcast — and that’s going to mean being more deliberate about how I structure my professional life to make room for them. Watch this space.
The previous post in this “STEM Warrior” series ended with an Elon Musk quote and a rallying cry to get back to work. I’m going to quietly retire the Musk fanboy part of my identity — the man turned out to be a more complicated figure than his Hyperloop white paper suggested. I’ve recalibrated since. Carl Sagan‘s open-minded scepticism remains my north star — it always has. And Bill Nye‘s proof that a mechanical engineer can pivot from Boeing to becoming one of the world’s most effective science communicators — without ever losing the engineer’s instinct for getting things right — feels more relevant to my own path than ever. More recently, Diamandis’s action-oriented optimism — ‘the best way to predict the future is to create it yourself‘ — has become an increasingly important compass during the post-ChatGPT era. And Jensen Huang’s vision — that AI will become as foundational as electricity, that everything that moves will eventually be robotic, and that computing can compress years of scientific progress into days — is something I’ve only recently started exploring, but his ‘speed of light‘ philosophy already feels like a worthy successor to the first principles thinking I used to associate with a certain other tech visionary.
The cause, though? That hasn’t changed. Our species needs — and deserves — a citizenry with minds wide awake and a basic understanding of how the world works. That’s still the Sagan quote on my “About Me” page, and it’s still the mission.
The tools are better now. The fork is real, even if the timelines are debatable. Time to re-tighten the bandana and figure out which path leads somewhere worth going. And hopefully, if AI progress continues on the trajectory that Shumer and Diamandis claim, this article will go out of date much faster than my rose-tinted piece about the Musk ecosystem did 🙂
Are you on the creator side of Fork 1? Or are you still waiting to see how it plays out? What’s the tool you’ve been meaning to push into your real work? And do you think Shumer is overstating things, or do you believe “The Singularity” is closer than we think? Leave a comment below.
Acknowledgements: this post was co-written with Claude Pro, and the images were generated in the free version of ChatGPT.
Leave a Reply