The AI Gold Rush: Lessons from the Third Industrial Revolution

Lessons from the Internet Revolution

I’ve got an idea that will make millions! And I’ll give you 50% if you build it for me.

There have now been four industrial revolutions:

• The first Industrial Revolution
• The Second Industrial or Technological Revolution
• The Third Industrial Revolution or Information Age
• The Fourth Industrial Revolution, or Industry 4.0

The Fourth Industrial Revolution is taking place right now, and AI is at the heart of it.

I was at university during the Technological Revolution, which is where I first encountered the internet. I was utterly fascinated by it. I finished my degree in physics and became a web designer and software developer instead of a scientist. Much more exciting!

Hardly anyone knew what “The Web” was back then - let alone what a Web Designer did. But as the internet boom erupted, suddenly everybody wanted to know. As soon as some millionaire-wannabe realised that there was money to be made - they just needed a sucker with my skills to create it for them - they didn’t hesitate to approach me with their idea for a million-dollar website.

It was the same offer every time - “I’ve got an idea that will make millions! And I’ll give you 50% if you build it for me.”

I grew tired of these ridiculous offers where I do all the work in exchange for being let-in on their idea. As if just coming up up with an idea would actually be of any worth to anyone.

As John Doerr once said, “Ideas are easy. Execution is everything.”.

The Proliferation of AI Systems

During the internet revolution, ideas were “ten-a-penny”. During the AI revolution they’re “ten thousand-a-penny”.

And now, with AI reaching the peak of its hype, everyone has got an idea.¹

Recently I attended an AI event for UK IT Leaders, and lots of ideas were presented - SaaS companies were building AI into products, consultancies were building agents, and the “hyperscalers” were selling chatbots.

After talking to the other attendees, I had a realisation: if ideas were “ten-a-penny” during the internet revolution, during the AI revolution they’re “ten thousand-a-penny”

And unlike during the internet revolution, you don’t need a particular set of skills to create anything using AI. Anyone can now build an AI app. Everyone can execute their ideas and make them real.

95% of AI Projects Fail

The result of everyone being able to create AI products is that every product is now AI-enhanced. Every tool is now AI-powered. Every IT project is now AI-enabled.

And yet most of it doesn’t work!

According to Fortune, a new report by MIT’s NANDA initiative showed that only 5% of AI pilot programs achieve rapid revenue acceleration. The vast majority fail to deliver any measurable impact on P&L at all.

You only have to try asking your favourite GenAI tool to create a Powerpoint slide with your company branding, or a cartoon image with some text overlaid on it, or give it unfettered access to your code to understand what failure to deliver might mean.

The Gold-Rush Mindset: Hype, Shortcuts and Hacking

Compounding the failure to deliver any real value, the rush to build something quickly can also lead to high levels of risk.

During those early internet years, the idea of “Gold-rush software development” had a profound impact on the way I approached my professional career.

A message which still feels prophetic today, McConnell warned that the habits of a gold rush — hacking, shortcuts, sleepless enthusiasm — rarely build lasting value.

Instead, the rush for gold leads to practices which are unfit for professional use. It leads to hacking.

As it was at the start of the internet boom, so it is for AI now. Uncontrolled, poorly governed AI development is a high-risk activity. Allowing anyone with “the right mix of innovation and initiative to pick it up” puts us at a high risk of failure.

It’s “AI hacking”.

Should We Even Use AI?

AI looks so promising, but if it fails to deliver so much of the time, wouldn’t it be sensible to ignore the gold rush altogether and simply disallow AI from being used in our organisations at all?

If the risks are so high, and controlling those risks is so hard, should we even bother with AI at all?

In the next article we’ll discuss why it’s no longer a question as to whether we should use AI, but only how you do it safely and strategically.