Spotting AI hype: telling substance from show

My LinkedIn feed is full of AI experts. Precise percentages, an uncomfortable truth, then a finished tool and a call to drop a codeword in the comments. It sounds like insider knowledge. Most of the time it is a template that thousands reuse. The good news: the tricks repeat, and once you see the pattern, you spot it everywhere.

I look at this as someone who audits rather than sells. What an auditor learns: the polished act says little. The small details say everything. Here are the patterns that separate well-packaged hot air from real substance.

Invented numbers with decimal places

The first warning sign is numbers that are too precise. If someone writes that a certain behaviour triggers exactly 15.6 percent more of something, pay attention. Decimal places are meant to sound serious. They are almost never backed by evidence. Real numbers come with a source and context, or they do not come at all.

As an auditor I follow a simple rule: a verifiable number or no number. A claim without checkable evidence is not a fact, it is an opinion in costume. Ask politely for the source. Substance gives you an answer. Show gives you an evasion.

Help or bait?

The second pattern is the bait. Drop a codeword in the comments and I will send you the PDF. It boosts reach in the short term, but it is a mechanic, not a gift. You type a word, the post gets interaction, and you land in a sales funnel.

Real help is openly accessible. Anyone who genuinely wants to give you something links the article, shows the example, names the source. Anyone who first routes you through a comment and a direct message is optimising their funnel, not your knowledge. It is not forbidden, but you should see it for what it is.

Substance is shared openly. Show is kept scarce and behind a price.

Selling a tool instead of understanding the process

The third and most important sign shows up in conversation. A credible advisor asks about your process first. A pretender talks about their tool immediately. Anyone who wants to build you an AI agent in the first meeting, without understanding your workflow, is not selling you a tool, but a risk.

Take three typical mid-market cases. An AI that pre-sorts applications. A camera with a model that inspects parts in final control. An AI that creates orders in the ERP. In all three the technology is the easy part. The hard part is the process behind it: who decides, who checks, what happens when it fails. Put AI on a broken process and it just automates the mess, faster and more expensively.

The self-appointed AI elite

The fourth pattern is the self-appointed elite. Lots of stage, lots of webinars, big words about the future. The uncomfortable question is: what has this person actually built, implemented or been accountable for? Anyone who never stood on the shop floor and never paid for their own mistake talks about AI the way people talk about the weather.

I have run audits in five industries, from aviation to precision manufacturing, in five countries, with over 1,200 documented audit hours. That does not make me the smartest person in the room. But I know what it is like when a process stops at half past three in the morning because a model fails to recognise something. You can hear that grounding in a person. You can also hear when it is missing.

Buzzword bingo as a substitute for substance

The fifth sign is the language. If a sentence collapses once you remove the fashionable words, there was nothing in it. Run the test: cut forward-looking, holistic, scalable and synergy from a consultant sentence. Is there a concrete statement left, or only hot air?

Substance sounds unspectacular. It names the part, the process, the number, the limit. It also says what does not work. That admission of limits is itself a sign of quality. Anyone who tells you AI solves everything has either understood nothing or wants to sell you something.

Your hype filter in five questions

You do not need to be an AI expert to separate the wheat from the chaff. These five questions are enough before you listen to someone, let alone sign a contract.

• Are the numbers backed by a source and context, or just made precise?
• Is the help openly accessible, or tied to a bait and a direct message?
• Does the person ask about my process first, or about their tool first?
• What has this person actually built, implemented or been accountable for?
• Is there a concrete statement left once I delete the buzzwords?

Anyone who answers these confidently and concretely has substance. Anyone who dodges, shows slides and talks about the big future without letting you touch a single example does not. It is no guarantee, but a reliable first filter.

And yes, the same principle shows up in the smallest detail. Anyone who does not even keep their own legal notice up to date is not checking their AI output either. Hype is loud. Substance is quiet and verifiable. When in doubt, trust the quiet.