The AI Revolution

Era One — Early Years

From Johnny 5 to GPT-2: A Lifelong Obsession

My relationship with artificial intelligence didn't begin in a boardroom or a university lecture theatre — it began in the living room, watching Short Circuit, mesmerised by Johnny 5, and later by Mr. Data in Star Trek: The Next Generation. There was something about the idea of a machine that could reason, feel, and grow that I found genuinely thrilling rather than frightening. I didn't just want to watch robots — I wanted to talk to one.

That instinct eventually pushed me toward programming. Long before AI assistants became a mainstream conversation, I was learning Python and putting it to practical use. At iSat Ltd in Farnborough, I designed the front-end interface for a multimillion-pound satellite tracking feed system developed in partnership with the European Space Agency — a project that demanded precision, reliability, and a solid grasp of how software could serve complex real-world systems. Alongside that, I was building automation scripts to eliminate repetitive tasks and sharpen workflow efficiency; the kind of work that makes a measurable difference quietly, without fanfare.

When large language models began to emerge, I was paying close attention. I was an early adopter of GPT-2 — not just as a curiosity, but as a working tool — experimenting with it for content generation at a point when most people had never heard the phrase "language model." Getting usable output from those early systems required patience, prompt intuition, and a genuine understanding of what the model could and couldn't do. It was rough, inconsistent, and occasionally baffling — and I was hooked.

"Technology is morally neutral until we apply it." — William Gibson

Era Two — The Modern Era

Working at the Frontier: LLMs, Agents, and Responsible AI

The leap from GPT-2 to modern large language models felt, from the inside, less like evolution and more like a phase change. I've worked extensively with Claude Code and ChatGPT Codex for software development — not just prompting for snippets, but using agentic coding workflows where the model plans, reasons through problems, scaffolds files, and iterates toward a working solution. Watching a command prompt think through a project structure in real time remains one of the most genuinely surprising experiences I've had in twenty years of working with technology.

Beyond code, I've built hands-on experience across the broader AI creative stack: image generation tools including Fooocus and Adobe Firefly; music generation with Udio and Suno; video production via Synthesia; voice synthesis through ElevenLabs; and agentic project management systems including Google's Jules. My YouTube channels have accumulated almost 100 million views, much of that content informed or supported by AI tools at various stages of production. I've released over 300 music tracks, narrated audiobooks to Audible standards, and co-authored content that spans multiple formats and audiences.

One project worth highlighting specifically: using Google Jules and ChatGPT in tandem, I managed the research, structure, and drafting of a 42,000-word manuscript exploring the intersections of social media, dopamine, and online marketing. That document now forms the foundation of a book in development — and it was built using agentic AI systems working toward a defined long-term goal, not just answering one-off queries.

On Responsible AI Use

I want to be clear about something, because it matters: AI is not a replacement for human creativity, judgement, or expertise. It is a catalyst. The difference is significant. A catalyst accelerates and enables — it doesn't substitute for the person directing the reaction. In my own practice, I cross-reference outputs across multiple models, using one to interrogate and stress-test the conclusions of another. I treat AI the same way I'd treat a very fast, very well-read research assistant who occasionally hallucinates citations: useful, impressive, and in need of oversight.

Aligning your question to your actual objective is a discipline in itself. Poor prompting produces confident-sounding nonsense. Rigorous prompting — knowing what you're asking for, why, and how to verify the answer — is a genuine skill, and one that separates practitioners from passengers in this space. The goal is always human-directed output that serves a real purpose, not volume for its own sake.

"AI isn't here to replace us, but to empower us." — Ginni Rometty

Era Three — What Comes Next

2026 and Beyond: AGI, Agents, and the Embodied Machine

We are, almost certainly, at an inflection point. The systems I use today — Claude, GPT, Gemini, Jules — are capable in ways that would have seemed implausible five years ago, and they are improving on a timescale that makes annual benchmarks feel quaint. The next wave isn't just smarter models; it's the convergence of intelligence with action. Agentic AI that can plan across days, manage projects autonomously, and interface with the physical world through robotics isn't a speculative future. It's a near-term engineering challenge that several serious organisations are actively solving.

Multimodal models — systems that reason fluidly across text, image, audio, and code — are already reshaping what it means to produce content, software, and knowledge. Tools like Claude Code represent an early glimpse of what happens when an intelligent system is given not just a conversation but an environment to operate in. The progression toward artificial general intelligence, whatever form that ultimately takes, will not be a discrete moment. It will be a gradient — and we're already on it.

As someone who grew up wanting to talk to a robot, and who has spent decades working at the intersection of technology and creativity, I find the current moment less frightening than fascinating. The questions worth asking aren't "will AI change everything?" — it already has — but rather: who is shaping it, toward what ends, and are the people building these systems as thoughtful about the implications as they are brilliant about the engineering?

That's a question worth staying curious about. I intend to.