{"id":1860,"date":"2026-10-12T09:00:00","date_gmt":"2026-10-12T07:00:00","guid":{"rendered":"https:\/\/matvision.eu\/the-certainty-of-uncertainty\/"},"modified":"2026-10-12T09:00:00","modified_gmt":"2026-10-12T07:00:00","slug":"the-certainty-of-uncertainty","status":"publish","type":"post","link":"https:\/\/matvision.eu\/de\/news\/the-certainty-of-uncertainty\/","title":{"rendered":"The Certainty of Uncertainty: When AI Models Are Confidently Wrong"},"content":{"rendered":"

Why AI models can be confidently wrong when sorting waste, and what that means for reliable recycling.<\/em><\/p>\n\n\n\n

By now, most of us have learned to treat the answers of generative AI with a healthy dose of skepticism. We have all seen ChatGPT invent a reference, a fact, or a place with complete assurance. In July 2026, two hikers in the Tatra Mountains had to be rescued by helicopter after following a \u201cshortcut\u201d suggested by ChatGPT: the route confidently led them onto climbing terrain they could not cross. The model did not say \u201cI am not sure.\u201d It simply answered, convincingly, and wrongly.<\/p>\n\n\n\n

For the general public, this skepticism towards AI predictions is relatively recent. At Matvision, however, it has always been the central engineering reality. Working with artificial intelligence on a high-speed sorting line requires a permanent, critical assessment of what the model predicts. More importantly, it requires engineering a system that remains reliable precisely where a standalone classifier would otherwise be confidently wrong.<\/p>\n\n\n\n

The premise is straightforward: a confidence score is not a probability. That gap is measurable, it is correctable, and a system that knows when to say nothing is worth far more than one that always answers.<\/p>\n\n\n\n

\n\n\n\n

1. The confidence score is not what it looks like<\/h2>\n\n\n\n

Measurement and material-characterization systems are standard across modern recycling. Among the most widely used are handheld XRF and LIBS analyzers, which give an immediate indication of a material\u2019s composition and suggest a corresponding grade directly in the field.<\/p>\n\n\n\n

But these devices are calibrated for specific families of materials. Their algorithms compare a measurement against a predefined database of reference grades, and in most cases the instrument is programmed to always return an answer<\/em>, even when the piece falls outside the scope of its calibration.<\/p>\n\n\n\n

A sorting system\u2019s neural network behaves the same way, but at a larger scale. It may report a confidence score of 0.95 for stainless steel grade 304, which an operator intuitively reads as: \u201cthis piece has a 95% probability of being 304.\u201d<\/p>\n\n\n\n

The mathematical reality is different. Modern classifiers use a softmax function, which converts the network\u2019s raw outputs into a set of values that sum to 1.0:<\/p>\n\n\n\n

P<\/mi>(<\/mo>class<\/mtext>i<\/mi>)<\/mo>=<\/mo>exp<\/mi>(<\/mo>z<\/mi>i<\/mi><\/msub>)<\/mo><\/mrow>\u2211<\/mo>j<\/mi>=<\/mo>1<\/mn><\/mrow>K<\/mi><\/munderover>exp<\/mi>(<\/mo>z<\/mi>j<\/mi><\/msub>)<\/mo><\/mrow><\/mfrac><\/math><\/div>\n\n\n\n

The reported confidence score is simply the maximum value within that normalized distribution. Because the values are constrained between 0 and 1, they look like probabilities. In reality, they express nothing more than a relative ranking among the specific classes the model knows. The softmax operation forces this ranking whether or not the true material is one of them.<\/p>\n\n\n\n

Consider a simple analogy: take a model trained only to differentiate cats from dogs, and show it an image of a giraffe. The network will not hesitate. It may answer \u201cdog\u201d at 99% confidence. What we would expect from a sensible model is something near a 50\/50 split meaning \u201cI have never seen this\u201d; that is exactly what a standard softmax cannot produce. It must distribute all its probability mass over the classes it knows, and it does so with manufactured certainty.<\/p>\n\n\n\n

On the line, the same thing happens. When a fragment belongs to an alloy grade absent from the training set, the model reports a high value for the closest known class, and the operator reads it as a probability of being that grade. This illusion of certainty is dangerous. Once an operator discovers it, trust in the system collapses. Not because the model is inaccurate, but because its confidence was never trustworthy to begin with.<\/p>\n\n\n\n

\n\n\n\n

2. Miscalibration is measurable<\/h2>\n\n\n\n

Stating that confidence does not equal probability is straightforward. The practical question is whether that gap can be measured systematically. It can.<\/p>\n\n\n\n

A model is considered well-calibrated when its stated confidence matches its observed accuracy. Over a large sample set, among all predictions made at 70% confidence, roughly 70% should be correct. Weather forecasting provides a clear operational parallel: if a meteorologist forecasts a 10% chance of rain across 100 days, and it rains on 50 of them, the forecast is not merely inaccurate; it is miscalibrated. The system systematically misrepresents its own uncertainty.<\/p>\n\n\n\n

That example is worth keeping in mind, because miscalibration runs in two directions:<\/p>\n\n\n\n