In this series, AI Start — Understanding AI without a technical background, we explore step by step what AI is, how it works in everyday work processes and how professionals can use it consciously and responsibly.
“We use AI.” You hear it more often, but it says little. AI can sit in two very different places in software, and the two are constantly confused. This article explains the distinction you need — and which question really matters to you as a user.
See
Two questions about software
Two questions help untangle “we use AI”:
- Was AI used to build the software?
- Is there AI inside the software when you use it?
These are separate questions. The answer to one says nothing about the other.
The first concerns the building process. A developer can use AI to write code faster. The user does not notice.
The second concerns the product itself. Does the software use AI while you work with it? Does it recognise something, summarise something or help you think?
Building and using are not the same.
Do
An app for swimming-pool water
Take a simple example: an app that assesses pool water. You enter the measurements — pH, chlorine — and the app tells you whether the water is good.
Now the two questions and four possible answers.
No AI to build it, no AI in use. A developer builds the app manually. The app compares your measurements with fixed target values. Nothing more.
AI to build it, no AI in use. The same app, but the developer had AI help write the code. Built faster. Exactly the same for you as a user.
No AI to build it, but AI in use. The app was built manually, but uses AI during its work.
AI to build it and AI in use. Both: AI helped build it, and the app uses AI while you work with it.
All four exist. That shows the two questions are independent.
What does the AI in the app actually do?
In the two cases without AI in use, the app follows fixed rules: is a value too high or too low, yes or no?
In the two cases with AI in use, this can mean:
- Reading a photo. You photograph a test strip. The app recognises the colours and converts them into values without you typing them in.
- Understanding a note. The operator types: “water looked milky this morning, busy weekend”. The app extracts important signals — cloudy water, heavy use.
- Weighing signals together. Each measurement looks fine on its own, but together they suggest a problem: combined chlorine is rising while visitors complain of stinging eyes. The app notices the connection and warns you.
- Explaining. Instead of just “pH 7.9”, the app explains what is happening and what you can do.
The difference from fixed rules is that the app is doing more than following target values: it interprets. That is what AI adds.
Own
The question that really matters
For the person using the app, one of the two questions matters: is there AI inside the software?
Whether AI helped build it is invisible. It does not change what the app does. The app is not smarter because it was made with AI.
Only when AI sits inside the software — recognising, interpreting, helping decide — does something change. Then you can expect different things. And you need to check different things.
So when someone says “we use AI”, the useful follow-up is simple: do you mean AI helped make it, or that the software itself uses AI?
Two different things. Two different conversations.
Summary
AI can appear in two places in software: in the building process and in the product itself. These are separate questions. Software built with AI can work entirely without AI in use, and vice versa. For the software user, what matters is whether the product contains AI. That changes what you can expect from the software and what you need to check.