Nicole Junkermann on Telling AI Substance from AI Hype

Why hype is the default setting

It helps to understand why hype is so common before trying to see past it. Artificial intelligence is genuinely impressive, which makes it easy to oversell. A striking demonstration can travel quickly, and excitement tends to outrun evidence. Add the incentives of attention and funding, and the result is a field where bold claims are the default and careful ones have to work harder to be heard.

Nicole Junkermann does not treat this as cause for cynicism. Some of the excitement is justified, and dismissing everything is as lazy as believing everything. The goal is discernment, not skepticism for its own sake. The interesting work is separating the advances that will still matter in a few years from the ones that will be forgotten, and that requires a steady method rather than a mood.

That method starts with a simple discipline: look past the demonstration. A demonstration is designed to impress, and it usually does. The harder and more useful question is what happens when the same system has to work reliably, at scale, for real customers who did not help build it. That is where substance and hype begin to separate.

Reliability matters more than capability

One of the clearest signals of substance is reliability. A system that can do something impressive once is interesting. A system that does it dependably, day after day, in conditions its makers did not control, is valuable. Nicole Junkermann pays close attention to this gap, because it is where many exciting technologies quietly fall apart.

Fluency is a particular trap. Artificial intelligence can produce answers that sound confident and authoritative while being wrong, and a polished surface can hide a shaky foundation. The question is not whether a system can generate a convincing response, but whether it is right often enough to be trusted for the task at hand. That is a different and far more demanding test.

This is why she values guests on the AI Overview who work close to deployment. They have learned that reliability is hard won, and they can describe the unglamorous work it takes to get there. When the conversation moves from what a system can do to what it does consistently, the hype tends to fall away and a clearer picture appears.

Follow the data, the cost and the customers

Three practical signals do a lot of work. The first is data. A genuinely strong product usually has a thoughtful answer to where its data comes from, how it stays current and why that gives it an advantage. Vague answers about data are often a sign that the substance is thinner than the pitch.

The second is cost. Impressive results sometimes depend on enormous expense that does not survive contact with the real world. A technology that is efficient enough for routine use has very different prospects from one that is only viable as a showcase. Asking what it costs to run a system at scale quickly separates the practical from the theatrical.

The third, and perhaps the most revealing, is real customer use. Are people who did not build the product paying for it, relying on it and coming back? Genuine adoption by independent users is the hardest signal to fake. Nicole Junkermann treats it as one of the strongest indicators that an advance is real, because customers vote with their time and their money, not their enthusiasm.

Listen to how people talk about limits

There is a subtler signal in the way people describe their own work. Those with real substance tend to be honest about limits. They can tell you where their system struggles, what it cannot yet do and what they are still working on. That candour is a mark of confidence and understanding, not weakness.

By contrast, a pitch that claims a technology can do almost anything, with no honest account of its limits, is usually a warning sign. Every serious tool has boundaries, and the people who understand a technology best are the ones who can describe those boundaries clearly. Nicole Junkermann listens for this carefully, because it reveals whether someone truly understands what they have built.

This is also why she finds the most useful conversations are the calm ones. A guest who is comfortable saying that something is hard, uncertain or still unproven is usually closer to the truth than one who promises that everything is solved. Honesty about limits is one of the clearest fingerprints of substance.

A grounded way to see the field

Put together, these signals form a grounded way to look at artificial intelligence. Look past the demonstration to reliability. Follow the data, the cost and the real customers. Listen to how people talk about limits. None of this requires deep technical expertise. It requires patience and a willingness to ask plain questions until the substance becomes clear.

This is the same discipline Nicole Junkermann brings to her work as the founder of NJF Holdings and its venture capital arm, NJF Capital, where telling durable progress from passing excitement is part of the job. It is also the spirit of the AI Overview. The aim is not to be the loudest voice in a noisy field, but to be a clear one, helping people understand what is genuinely changing and what is merely being announced.