The Hype Is Not the Technology

I work with AI nearly every day. I build practical systems with it for small and medium businesses, I have spent 25 years delivering enterprise technology across banking, media, automotive and government, and I am, on the whole, an optimist about what this generation of tools can do. People sometimes assume this makes me a promoter of the current excitement. The opposite is closer to the truth. It is precisely because the technology is real that I think the hype surrounding it should be treated as a problem in its own right: a tax on the attention of every business trying to make a sound decision right now.

This essay makes one argument. The AI cycle we are living through has shifted attention from solving business problems to performing innovation, and that shift has two especially corrosive expressions: AI as a story for raising capital, and AI as a cover story for layoffs. Both are well documented. Both are rational responses to incentives. And both carry the same hidden cost, which is that businesses lose hold of the only questions that ever made technology pay.

Before going further, three clarifications, because essays like this are easily misread. First, I am not arguing that AI is overrated; in many narrow, well chosen applications it is genuinely underrated. Second, I am not claiming that every workforce decision attributed to AI is dishonest; some roles really are changing. Third, I am not advising anyone to wait; sitting still has costs too. The argument is narrower and, I think, more useful: the reasons most loudly given for adopting AI are increasingly disconnected from the reasons a business should adopt anything.

We have been here before, except we have not

Every technology cycle of my working life has followed the same arc. A genuine capability appears. A narrative forms around it. Capital chases the narrative rather than the capability. The sober operators who keep asking ordinary questions (what does it cost, what does it replace, who maintains it) get talked over for a few years, and then quietly turn out to have been right about most things.

The dot-com years gave us companies whose share prices moved simply because they added ".com" to their names. The blockchain wave produced the spectacle of an iced tea company rebranding itself around distributed ledgers and watching its share price nearly triple in a day. Big data, and then the metaverse, ran smaller versions of the same play.

So far, so familiar. Here is what I think is different this time, and it is the opposite of reassuring. In previous cycles, the gap between story and substance eventually became obvious to everyone. With AI, the underlying capability is real and improving quickly, and that makes the hype more dangerous rather than less. Counterfeit money circulates most easily in an economy where the real currency holds value. When some AI claims are true, every AI claim borrows credibility it has not earned, and the work of telling them apart lands on buyers who never asked for that job.

The first misuse: capital theatre

Watch how companies describe themselves to investors and you can see the incentive at work. Pitch decks that were about logistics, lending or rostering eighteen months ago have been rewritten so that AI is the subject of every sentence. The substance underneath has often changed far less than the language. Founders are not foolish; they are responding to a market that, for the moment, pays more for story velocity than for problem clarity. Capital raised on a narrative is still capital, and in the short term the strategy works, which is exactly why it spreads.

It has spread far enough to earn a name. Regulators in the United States have already fined investment firms for overstating their use of AI, and Australia's corporate regulator has put companies on notice about the same behaviour. "AI washing" now sits beside "greenwashing" in the enforcement vocabulary, and when a practice acquires its own portmanteau and its own enforcement actions, it is no longer an edge case.

For the people I work with, owners and executives of small and medium businesses, the cost of all this is practical rather than abstract. When every vendor claims AI, the claim stops carrying information. The signal a buyer actually needs (will this remove a constraint in my business, at a cost I understand) is buried under language designed for a different audience entirely: investors, analysts and headlines.

The second misuse: the layoff story

The second misuse is quieter and, I think, more corrosive. Between 2020 and 2022, much of the technology sector hired as though pandemic-era demand curves and near-zero interest rates were permanent. They were not. Demand normalised, money became expensive, and the over-hiring had to be unwound. That is an ordinary, if painful, business correction, and there would be no shame in describing it plainly.

Instead, a remarkable number of announcements described it as AI-driven efficiency. The phrasing is well chosen: markets tend to punish "we hired for a world that did not arrive" and reward "we are at the frontier of automation". I understand the temptation. I still think it is a mistake, for three reasons that have nothing to do with sentiment.

It blames a technology for a forecasting decision, which corrodes trust inside the company, because the people who remain can usually tell the difference. It writes a cheque the technology must now cash: if the announcement says AI replaced a team, the systems had better perform like the team, and in most cases today they cannot, not yet, not unattended. And it teaches the public that AI adoption equals headcount reduction, which makes every future, legitimate AI initiative harder to lead, because staff have learnt to hear "efficiency" as "redundancy". It is telling that some of the companies that announced AI had replaced hundreds of roles have since, quietly, begun hiring people back.

What the optimists get right

None of this works as an argument unless I concede what the optimists get right, so let me do that properly. The capability curve is steep, and anyone extrapolating from what these systems could do two years ago is reasoning from stale data. In narrow, well instrumented tasks (drafting, classification, summarisation, first-pass analysis of documents) the productivity gains are real and measurable today. Some roles genuinely are being reshaped. And waiting carries its own price: the businesses that learn these tools on small problems now will move faster on large ones later.

All of that is true, and none of it rescues the behaviour described above, because the argument was never against AI. It is against adopting anything for reasons that have nothing to do with your business. The technology being real does not make your reasons good.

The questions that never changed

Strip the noise away and the questions a business should ask about AI are the questions it should always have asked about technology. What problem are we solving, and for which customer? What outcome will change, and how will we measure it? What is the full cost of the change: not the licence, but the integration, the training, the months of reduced productivity while new habits form, the maintenance that never appears on the slide? Who is accountable for the result, by name? And what happens after go-live, when the consultants are gone and the system meets an ordinary Tuesday afternoon?

I have watched the same meeting happen across three decades; only the noun changes. Mainframe, ERP, cloud, data, now AI. The companies that did well out of each wave were rarely the loudest adopters. They were the ones who understood their own processes and their own data before automating either, who counted the cost of change honestly, and who retained the most underrated capability in management: the ability to say no, or not yet, without feeling left behind.

Technology fails at the human layer far more often than at the technical one. In 25 years I have seen very few systems fail because the software could not do what the brochure promised. I have seen a great many fail because nobody changed how the work was actually done. AI does not repeal that rule. If anything, tools that produce plausible output faster than people can verify it raise the premium on process discipline rather than lowering it.

Why I built my company the way I did

This is the point where I should declare an interest, because I have built a company on these convictions, and I would like you to know how, since the structure is the argument.

Company31 is an independent change and digital transformation advisory for small and medium businesses across Australia and New Zealand. Independent means exactly that: we do not resell software, we take no vendor commissions, and no platform pays us to recommend it, so when we tell a client a tool is right for them, the sentence contains no hidden second clause. We tie our fees to outcomes rather than effort, because advisors who are paid for time have a quiet incentive to let change run long. And we run every engagement on an artefact we call the Commitment Ledger: a plain record of who promised which outcome, by when, at what cost, reviewed in the open until it is delivered or honestly renegotiated. It is not glamorous. It is the most clarifying document I have used in a quarter century of delivery.

We compress the philosophy into six words: see clearly, choose well, change once. If the AI era rewards anything, my guess is that it will reward exactly that.

A filter you can use on Monday

If you lead a business and the pressure to "do something with AI" has reached your desk, here are the five questions I would put to any proposal before a dollar moves:

1. What constraint in our business does this remove, and how will we measure that it is gone? If the honest answer is a competitor's name, or "everyone is doing it", stop.
2. Could we describe the value without using the word AI? If the benefit only exists inside the vocabulary, it may not exist at all.
3. Do we understand the process and the data underneath it today? Automating a process you do not understand gives you the same mess, faster.
4. What is the full cost of change, including adoption, integration, training and maintenance, not merely the subscription?
5. Who owns the outcome by name, and what happens after go-live? None of these questions require technical knowledge. All of them require a willingness to look unfashionable in a meeting for about ninety seconds. That is the entire price of clarity at the moment, and it is the best trade available.

Chop wood, carry water

There is an old Zen saying: before enlightenment, chop wood, carry water; after enlightenment, chop wood, carry water. The state of mind changes; the work remains.

AI will change a great deal, and I am glad to be building with it. But after AI, you will still need to know your customer, your process and your numbers. You will still need people who change how they work, not merely what they click. The hype will pass; it always does, and usually faster than its loudest voices expect. What remains afterwards is what was always there: businesses that see clearly, choose well and change once, compounding quietly while the noise plays itself out.

The technology was never the point. It was always the enabler. The point is, and remains, your business.

---

Manuel Re is the founder and Director of Advisory Services at Company31, an independent change and digital transformation advisory serving small and medium businesses across Australia and New Zealand. company31.com