So often noise and emotion drive investment decisions, rather than underlying fundamentals. But as any bona fide professional overseeing a long-term investment portfolio will tell you, that’s a race to the bottom. 

AI is subject to plenty of noise. Every time the NASDAQ wobbles, a familiar argument returns: AI is just a bubble, productivity gains are imaginary and investors have got way ahead of themselves.

That’s missing the point. AI has clearly attracted enormous enthusiasm. That’s what happens when markets encounter something they haven’t quite yet seen before. None of that actually tells us whether AI itself is overhyped, or actually unpacks its underlying long-term investment merits.  

What we do know is that the technology works. One only has to play around with it for a few minutes to realise just how powerful it is. 

From an investment standpoint the simplest AI comparison I keep returning to is electricity. When electricity first arrived at scale in the early 20^th century, it was used in some pretty obvious ways: better lighting, cleaner factories and more reliable power.

Incredibly useful improvements, certainly, but few people imagined assembly lines, refrigeration, elevators or entire industries being reorganised around electrification. Just like AI, the first-order effects were visible immediately, but the second-order effects changed everything.

The reality is most organisations are still using AI like that initial use of electricity.

It helps people write faster and more clearly, research faster, code faster and automate bits of existing work. But the more meaningful impact is likely to come when businesses across all industries start redesigning themselves around the technology rather than simply bolting it onto existing workflows.

That could mean healthcare systems identifying risks earlier, small businesses accessing capabilities once reserved for large corporations, or manufacturers predicting equipment failures before they occur.

In healthcare, AI is already helping clinicians analyse scans, identify patterns in patient data and reduce administrative workloads. In drug discovery, researchers can test and model potential treatments far more quickly than was previously possible. In manufacturing and logistics, businesses are starting to optimise supply chains, reduce downtime and make better decisions using real-time data.

Education may prove another significant example. AI-powered tutoring tools can provide personalised support tailored to an individual’s learning style and pace, allowing teachers to spend more time on coaching, critical thinking and student engagement.

As with electricity, the biggest gains may come not from doing existing tasks faster, but from enabling entirely new ways of working.

Those outcomes are difficult to fully imagine today, just as few people looked at an electric light bulb and predicted modern manufacturing, refrigeration or electric vehicles. The first wave of benefits from AI may be measured in minutes saved. The second wave could be measured in entirely new products, services and industries that have yet to emerge.

That is why demanding immediate, precisely measurable productivity gains from AI is misguided. For decades we have given employees computers, email, Excel, PowerPoint and Word without asking each person to prove the productivity uplift from each licence. We do it because the benefit is obvious once the tool becomes part of competent work. AI is moving into that same category.

Another lesson from history is that technological breakthroughs do not always translate into profits straight away. Railways, electricity and the internet all transformed economies long before they consistently transformed company earnings.

This gap between innovation and profits is often where investors get into trouble, and a space where the noise can become deafening, and distracting.  

Expectations often run ahead of reality, valuations can become stretched and some companies inevitably disappoint. But that does not mean the technology itself has failed. It simply means the path from invention to widespread commercial value is rarely linear.

There are understandable reasons why some people seem keen to declare the AI boom over. Some concerns are serious and deserve respect. Questions around market concentration, capital discipline, energy use, copyright, safety and employment disruption are not trivial issues.

But there is also a less noble temptation at work. It is intellectually satisfying to say the crowd is foolish. There is an allure to being the person who sees the bubble before everyone else. The appeal is even greater when the thing being dismissed is big, complicated, expensive and threatening to existing ways of working.

The problem for the AI sceptics is that betting against human ingenuity, science and progress has historically been a difficult way to make money. These forces compound small gains relentlessly in a way that is invisible from one day to the next but eventually overwhelms even the most persuasive bear case with the passing of years.

For long-term investors, that’s important to keep in mind.

Stuart Eliot is the general manager, investments at AMP