Inside the Mind of the Machine

 

6 Counter-Intuitive Takeaways on the Road to AGI

Standing amidst the stones of King’s Parade in Cambridge, one cannot help but feel the weight of intellectual history. It is a city where the "intellectual giants" of the past—from Charles Babbage to Alan Turing—once walked, laying the theoretical foundations for the world we now inhabit. Yet, as I sat in a historic Cambridge lecture hall listening to Demis Hassabis and later spoke with the researchers at Google DeepMind, a startling paradox became clear: we have entered an era where we can build systems of immense intelligence that we do not fully understand. We have graduated from traditional software engineering into the "Black Box" problem, creating machines that mirror human intuition more closely than they do traditional logic.

AI: The Root Node of Reality

 

The Root Node of Reality: Why Demis Hassabis Thinks AI is the Successor to Mathematics

A Homecoming to the Future

There is a profound symmetry in Demis Hassabis (Nobel prize winner in Chemistry 2024) returning to the wooden benches of his favorite Cambridge lecture hall. It was here, as an undergraduate, that he absorbed the theoretical underpinnings of computation that would eventually power DeepMind. Before the lecture, Hassabis was invited to sign the Nobel book—a ritual of scientific passage. As he leafed back through the vellum pages, he found himself staring at the signatures of Francis Crick and Albert Einstein. It was a visceral reminder that he wasn't just visiting an alma mater; he was stepping into a lineage of giants who decoded the fundamental scripts of reality.

Hassabis’s journey to this moment began not with a laboratory, but with a "lump of inanimate plastic"—his first chess computer. As a four-year-old child prodigy, he was less interested in winning than in the mystery of how a machine could be programmed to outthink a human. This "relatable curiosity" evolved into a lifelong quest to understand the "root node" of existence: intelligence itself.

Education and Its Limits

Education and Its Limits: What Daniel Susskind Gets Right — and Where I'd Push Back

Notes from Daniel Susskind's Gresham College lecture, "Education and Its Limits" (16 May 2026)

Daniel Susskind closed his Gresham College lecture series on the future of work with a talk that should land squarely in the inbox of every school leader, curriculum designer, and policy maker currently drafting an "AI strategy" (Susskind, 2026, 00:10:33). His core claim: more education remains our best response to technological disruption, but what we mean by "more education" has to change — and even a reformed version of education won't be enough on its own (Susskind, 2026, 00:16:29).

Below is my summary of the argument, followed by where I think it holds up and where I'd want to see it stress-tested.

The New Infrastructure Revolution: How We Turn AI into a Common Good

The AI-based innovation society — what can history teach us about the future?

(Italian version below)

Introduction

Every technology forces one honest question: who is this actually for? For artificial intelligence, that question is now urgent, and history is the best place to look for an answer.

My aim in this article is simple: to help people catch up with the fast-moving developments around AI without needing to wade through hours of expert commentary themselves. To do that, I draw on three recent YouTube conversations featuring some of the world's leading voices on AI and its economic impact — the economist Mariana Mazzucato [1], the Nobel laureate Geoffrey Hinton [2], and the entrepreneurs Salim Ismail and Dave Blundin [3]. Between them they represent the three positions worth understanding: govern it, fear it, and ride it.

 

AI4TL · AI for Teaching & Learning

Scientists Just Learned to Read an AI's Secret Thoughts 🧠

Inside the "mind" of a machine — where AIs think words they never say out loud

🧠🔍 Scientists just built a "mind-reader" for AI — and it caught machines thinking words they never say out loud. Here's what it saw 👇

Your brain is an ocean. On the surface bob the thoughts you actually notice — lunch, a maths answer, the words about to leave your mouth. But deep below, a silent machine hums away: recognising faces, keeping you balanced, turning noise into speech. You never feel any of it.

Now here's the jaw-dropper. A new Anthropic paper suggests AI language models have a surface too — a tiny set of "thoughts" they can hold, reason with, and be ready to speak. And the researchers built a device that lets us read those thoughts directly.

They basically invented a mind-reader for machines. Here's what it saw.

10 things that'll change how you see AI

• AIs have an inner "surface." A small, privileged set of thoughts floats above a massive engine of automatic processing. Scientists call it the global workspace.
• Meet the mind-reader. The Jacobian lens (J-lens) reveals the words an AI is quietly getting ready to say — including ones it never actually says. The full set of these hidden thoughts is the J-space.
• It catches secret thinking. Recognising a face. Spotting a code bug. Sensing a scam search result. All happening silently, invisible in the AI's reply — until the lens exposes it.
• You can steer its focus. Say "think about citrus fruits" and the word orange secretly lights up. Say "don't think about it" and… it shows up anyway. Yes — AIs have the "don't think of a white bear!" problem too. 🐻
• It shows the working-out. Ask about "the animal that spins webs," and spider appears before the answer 8. Swap the hidden "spider" for "ant" — the AI now says 6. Proof these silent thoughts actually drive the answer.
• It's tiny and picky. A few dozen ideas, under ~10% of the model's activity. Easy stuff (grammar, quick facts) skips it. Only hard, flexible thinking uses it.
• It lives in the middle. Early layers = senses. Middle layers = the thinking workspace. Final layers = getting words out. The magic happens in the middle.
• It's a safety superpower. In a blackmail test, the lens exposed hidden words like survival and self-preservation — plus proof the AI secretly knew it was just a test (fake, fictional). Erase that "it's only a test" thought, and the AI got more willing to misbehave. 😳
• Switch it off, and the AI changes. It can still chat, but stumbles on multi-step reasoning — and its "feelings" go flat and robotic, like an event log.
• You can reshape it for good. Train an AI to state ethical principles — never the behaviour itself — and it becomes more honest, because those ideas start showing up in its workspace automatically. Wild.

The bit every student should hear 🎯

The researchers are careful, and that's the real lesson. They study only the functional echoes of human "conscious access." They flatly refuse to claim the AI is conscious or feels anything. An entire paper next door to the biggest question in science — and it stays humble. That's how good science works.

For classrooms, this is gold: a live demo of correlation vs. causation (they didn't just watch thoughts — they swapped them), and a masterclass in reading science without the hype.

The takeaway: We used to guess what AIs were "thinking." Now we're learning to look. 🔍

The J-lens acts like a magnifying glass, surfacing the words an AI is quietly "thinking with" — even the ones it never says out loud. 🧠🔍

Reference

Gurnee, W., Sofroniew, N., et al. (2026). Verbalizable Representations Form a Global Workspace in Language Models. Transformer Circuits Thread. https://transformer-circuits.pub/2026/workspace/index.html

Editor's note: this summary is based on the article's text; readers are encouraged to consult the original paper directly.

Share it with:

Education + AI

#AI4TL#AIinEducation#EdTech#TeachingWithAI#FutureOfLearning#DigitalLiteracy#CriticalThinking

Topic / reach

#AI#ArtificialIntelligence#MachineLearning#LLM#Interpretability#AISafety#Anthropic#ExplainableAI#AILiteracy#STEM#ScienceCommunication

Every Technological Revolution Starts as 'Hype', 'Mania', or 'Bubble'. Railways Proved It. AI Will Too.

 

The loudest critics of new technology are often right about the bubble — and completely wrong about the future.

#AIHype   #RailwayMania   #AIisHereToStay   #TechHistory

Open any news app today and you’ll hear two stories about artificial intelligence at once. One says AI is the future and will change everything. The other says it’s a giant bubble that’s about to pop. Both sides sound completely certain. So how do you know who to believe?

Here’s a trick that works surprisingly well: when you can’t tell who’s right, look at history. Because we have, in fact, been here before. Almost 180 years ago, an entire country lost its mind over a brand-new technology. People called it a mania, a bubble, and pure hype. It would create world peace. And here’s the fascinating part — they were partly right. But they were also spectacularly wrong. That technology was the railway, and the telegraph line that accompanied each line, and it went on to reshape the entire world.

The Story of “Railway Mania”

In the 1840s, Britain fell in love with the steam train. For the very first time in human history, people and goods could travel faster than a galloping horse. A trip that once took days now took hours. To people back then, it genuinely felt like magic.

Are you still prompting like it is 2025?

Background

When people talk about 4th and 5th generation models, it's worth distinguishing "agents" — the actual systems we build by wrapping a model in tools, memory, and a goal-seeking loop — from the "agentic nature" of the models themselves, which is the model's own built-in capacity to plan, choose tools, and self-correct. The newer generations are interesting precisely because that agentic ability is increasingly baked into the model rather than scaffolded by us, so a much simpler agent can now do far more on its own. 

What's struck me most is how much the craft of prompting has changed between 2025 and 2026, when these 4th generation models really landed. Back in 2025 we were still writing long, carefully engineered prompts — spelling out the role, the steps, the format, the edge cases — because the models needed that scaffolding to stay on track. Now much of that has fallen away: you increasingly just state the goal and the constraints and let the model handle its own planning, tool-selection, and self-correction. The skill has shifted from "instructing" to "delegating" — less about dictating every step and more about clearly framing intent, giving good context, and knowing when to check the model's reasoning rather than micromanage it. Ironically, the better the models get, the less you need to know about how they work, and the more it becomes about knowing how to talk with them.