A Short History of AI — What It Does, What People Worry About, and Where It's Going

Artificial intelligence isn't new. The idea that machines could think has been around since the 1950s, when Alan Turing asked a deceptively simple question: could a machine ever convince a human it was also human? That question — the Turing Test — set the direction for the next seventy years of research.

The Quiet Decades

For most of that time, AI stayed a research topic, not a product. Universities worked on it. Governments funded it. It barely touched ordinary life. The progress was real but incremental — chess programs that slowly got better, spam filters that slowly got smarter, voice recognition that almost, but not quite, worked.

That changed around 2012, when a technique called deep learning started producing results that surprised even the people building it. Image recognition went from unreliable to superhuman in a few years. Translation went from laughable to genuinely useful. Voice assistants — Siri, Alexa, Google Assistant — went from novelty to something people actually relied on.

The ChatGPT Moment

In November 2022, OpenAI released ChatGPT. It reached 100 million users in two months — the fastest adoption of any consumer app at the time. Almost overnight, AI stopped being something you read about and became something anyone could open in a browser and use for free.

That moment forced every industry, trades included, to ask the same question: what does this actually mean for us?

What People Worry About

The honest answer is that some of the worry is justified. Three concerns come up more than any others.

Jobs. This is the one people feel most, and it's worth taking seriously rather than waving away. In 2016, one of the most respected researchers in the field — Geoffrey Hinton, now a Nobel laureate and often called the "Godfather of AI" — told the world:

"People should stop training radiologists now. It's just completely obvious that within five years deep learning is going to do better than radiologists."

He went further, telling radiologists they were "like the coyote that's already over the edge of the cliff but hasn't yet looked down."

It didn't happen. The opposite did. AI got dramatically better at reading X-rays and scans — genuinely faster and, in narrow tasks, more consistent than a human eye. But someone still had to interpret what those scans meant for a real patient, weigh it against their history, and decide what to do next. AI could generate the images faster than ever; only a radiologist could actually diagnose. Hospitals that adopted the technology found their radiologists processing far more scans per day, seeing more patients, and generating more revenue for the department — which meant they hired more radiologists to keep up, not fewer. Mayo Clinic's radiology headcount is up over 50% since Hinton's prediction (Radiology Business, on Hinton's New York Times comments). Radiologists are now among the best-paid specialists in medicine, and demand is still climbing. Hinton himself has since said he got the timing wrong, and that he spoke too broadly: he only meant reading the images.

That's not a one-off story about medicine. It's the pattern almost every real deployment of AI follows: the technology removes a bottleneck, the business does more of what it was already doing, and it needs more people to handle the extra volume — not fewer. It's uncomfortable for the specific person whose task gets automated, and that discomfort is real and shouldn't be dismissed. But "AI will make this profession obsolete" has a poor track record as a prediction, and "AI will change what this profession spends its time doing" has a very good one.

Accuracy. AI systems can be confidently wrong. They generate plausible-sounding text that isn't always factually correct. This is a real limitation, not a minor footnote, and anyone using AI for their business needs to understand it going in. It's exactly why we review everything our tools produce before it goes anywhere near a customer.

Privacy. AI systems learn from data, and legitimate questions exist about what data is used, how it's stored, and who can see it. We process data in line with UK GDPR; our privacy policy explains how.

What AI Actually Enables

For small businesses and tradespeople, AI isn't about replacing people. It's about handling the work nobody wants to do at 9pm:

• Answering customers' questions on your website at 9pm, when you're putting the kids to bed
• Booking them in straight into your diary, instead of a message to chase in the morning
• Texting back a missed call with a link, so the customer isn't left with silence
• Sending booking confirmations without anyone typing them by hand

None of this is a factory-floor robot replacing a tradesperson. It's a digital assistant handling the office work so the tradesperson can stay focused on the trade.

Growth

The global AI market was worth about $137 billion in 2022. It's projected to reach $1.8 trillion by 2030, growing around 37% a year, according to Grand View Research (AI Magazine on the report). That growth is coming from ordinary businesses adopting AI tools, not just from tech companies building them.

For a trade business, the real question was never whether AI would touch your trade. It's whether you'll be the one using it, or the one competing against someone who already is.

This post is part of Machine Room: the big picture of AI, explained without the jargon.