When you hear about "artificial intelligence", what do you picture? A robot that reasons like a human? A computer that feels emotions? An entity that is about to surpass us and make us obsolete? If your mental image resembles science fiction more than reality, you're not alone: that is exactly how AI is portrayed by most media, tech companies and even some researchers.
The truth is much more prosaic — and in some ways more interesting. Artificial intelligence, as it exists today, is not "intelligent" in the human sense. It does not think, does not understand, does not feel emotions, does not have its own goals. It is a set of mathematical and computational techniques that allow computers to recognise patterns in data and produce output based on those patterns. When an AI system "recognises" a face in a photo, it does not "see" the person: it calculates the probability that a certain set of pixels matches a pattern it has encountered millions of times in training data.
This distinction is not academic pedantry: it has enormous consequences for how we regulate, use and think about these technologies. If we believe AI "thinks", we might entrust it with decisions it should not make — like determining whether a prisoner deserves parole. If we believe AI is "just maths", we might underestimate its real impact on people's lives.
Scholar Kate Crawford uses an effective image: AI is neither intelligent nor artificial. It is not intelligent because it has no understanding; it is not artificial because it depends entirely on material resources (rare earths, energy, water) and human labour (data labellers, moderators, crowd workers). It is a socio-technical system, not an autonomous entity.
Understanding what AI really is constitutes the first step towards being able to talk sensibly about its impact on society. This is not about being "pro" or "against" AI — it is about having the tools to distinguish between what is real and what is marketing, between what is possible and what is desirable.
In the rest of this module, we will explore how these technologies actually work — from machine learning to Large Language Models — to build a solid understanding on which to base the conversations of the following modules.