WORDS —
TIM WARREN
Artificial intelligence is the phrase on everyone’s lips, sitting somewhere between a business buzzword and a source of genuine anxiety.
The story of AI isn’t magic; it is a history of maths, hardware, and two ways of thinking.
A HISTORY
The second world war essentially launched computing (as portrayed in the movie The Imitation Game, about Alan Turing), yet computing was a physical ordeal. Machines were giant, clunky beasts that ran on relays and valves, generating immense heat for modest calculations.
Then, in 1947, the transistor was invented – a tiny component orders of magnitude smaller and faster than anything preceding it, kickstarting the true age of computing.
In those early days, computer scientists had to decide how these machines should ‘think’. Most went down the path of Deterministic Computing: exact, logic-based systems where input A always leads to output B. This path dominated the market for 50 years, used in everything from phones to calculators.
But a few computer scientists experimented with Statistical Computing (or Probabilistic Computing), adopting probability over logic. For decades, this remained a niche academic pursuit that occasionally crept into our lives. Predictive text
on your old Nokia? That was a statistical model guessing the next word. The phone camera that detects a face? That isn’t ‘seeing’ a nose and eyes, it’s statistically recognising a face-shaped pattern. It was never exact. It was a best guess.
THE TRANSFORMER EXPLOSION
The tipping point arrived in 2017 when researchers published a paper on a concept called Transformers. It sounds like sci-fi, but it was essentially a method to build massive statistics models that could ‘learn’ from vast amounts of data to make highly accurate guesses. These approaches became known as Generative Pre-trained Transformers, or GPTs.
Key companies like OpenAI were paying attention. When ChatGPT was released in 2021, it didn’t just succeed, it became the fastest-adopted technology in history, outpacing the iPhone and the internet itself – gaining 100 million users in a month.
Quickly, Google, Meta, Anthropic and others followed suit with their own versions.
THE ‘TEENAGE INTERN’ DILEMMA
So, is it actually intelligent? The short answer: no. It is a synthesis of intelligence. Tools like ChatGPT or Gemini are Large Language Models (LLMs). They have ingested much of the sum of human knowledge from the likes of books and Wikipedia, and use that data to predict the answer that will make you happy.
Because they are probabilistic, not deterministic, they can be incredibly convincing while also being completely wrong! I suggest people treat AI like a smart, dramatic, unwise teenage intern.
Assume it means well, but, verify everything. If you ask it to translate English to German, it will do a remarkable job. If you ask it to write your wedding vows, it might succeed (though I don’t recommend it). But if you ask it to manage your wedding budget? Don’t. It doesn’t ‘know’ maths; it just predicts what a maths answer looks like.
NAVIGATING THE WAVE
We are now in a phase of global adjustment. Universities, schools and governments initially tried to ban these tools, but the consensus has shifted from prohibition to integration. The question is no longer if we use AI, but how we use it safely.
THE GOLDEN RULES OF AI SAFETY AND USE:
The privacy trap. Never put private data into a public AI. Whether it’s client strategy notes, credit card details, or family secrets. Once you type it in, it’s part of the model. I have seen data entered in one terminal leak out of another. Treat the text box like a public bulletin board.
The swimming analogy. The technology changes weekly. It is easy to feel overwhelmed. But think of it like the ocean: you don’t need to surf every wave. You just need to catch one and get comfortable.
Always learning. The pace of change in this area is extraordinary. I have worked in technology my entire career and I’ve never seen anything like it. Don’t imagine you can learn everything, just learn the things that help you in your day-to-day while remembering that you need to keep learning as the environment changes.
Edit relentlessly. AI is excellent at being an editor. For this article, I fed a rough 2,000-word draft into an AI and asked for structural feedback. It was a brilliant companion. But if I had asked it to write this from scratch? It would have been generic, flat, and likely factually incorrect.
WHAT IS IT GOING TO MEAN FOR OUR FUTURE?
Kids are going into very different roles to what parents went into. A core capability for any desk job will be the ability
to use artificial intelligence to enhance results. A handful of the occupations that will be fundamentally changed by artificial intelligence in the near term include law, software development, advertising, and marketing. There is a saying: “If you’re not worried about where this is going, you don’t know what’s going on!”
AI is not a replacement for human intelligence; it is a mirror of it: flawed, vast, and capable of great utility if held in the right hands.
THE TOOLKIT & THE FUTURE
While ChatGPT grabbed the headlines, the ecosystem is vast. The major players include Anthropic’s Claude, Google’s Gemini, and X’s Grok. These models are predominantly owned by tech giants because the training costs are astronomical.
Search vs chat
There is a growing trend to use artificial intelligence chat instead of internet search, but there’s actually a big difference:
- Search engines are indexes of existing resources. They give you a list of resources (links) so you can find the answer.
- LLMs (AI Chat) attempt to be the resource. They synthesise an answer directly.
- AI chat can be out of date and incorrect, but is so convinced that it is right it can be alarming!
- If you want to analyse Romeo and Juliet, a search engine gives you essays to read. An LLM writes the essay for you. Parents: talk to your kids about this. Using AI to cheat robs them of learning, but using it to brainstorm is a modern superpower.
Gaming
One of the most exciting, lower-risk arenas for AI is gaming and video. In a fictional world, ‘hallucinations’ (errors) aren’t fatal – they’re features. Developers are using AI to generate breathtaking visual scenes and dynamic storytelling that reacts to the player in real-time. This is where we will likely see the most dazzling, unbridled applications of the technology in the coming years.
Tim Warren studied computer science including artificial intelligence at Auckland University. In 2017 he founded AI company Ambit to automate customer service. He is a regular writer and presenter on artificial intelligence.
