AIorGenerative AI
Artificial intelligence is the name of an entire field, one that holds systems which classify, predict or sort without producing anything. Generative AI is a recent branch of it, the one that makes new content: text, image or sound.
Point by point
| Criterion | Floor 1Floor 1 · The Modela solid block on its own: the prediction machineAI | Floor 1Floor 1 · The Modela solid block on its own: the prediction machineGenerative AI |
|---|---|---|
| What the system produces | A decision, a score, a ranking | New content |
| Age | The field is seventy years old | In general use since 2022 |
| Everyday example | A spam filter, a route calculation | An assistant that drafts |
| How it is evaluated | A measurable error rate on known cases | Judgement on open-ended outputs |
| Where the risks are | Unfair decisions, discrimination | False content, infringement, leaks |
| Relation between the two | Holds the second | Is held within the first |
On the ground four situations
Your bank refuses a loan on the strength of a score computed by a system trained on past applications.
Floor 1 · The Modela solid block on its own: the prediction machineAINothing is produced here, a score is computed. This is AI in the sense of the field, and it is the kind of use that regulations frame most strictly.
A tool drafts the minutes of a meeting from its recording.
Floor 1 · The Modela solid block on its own: the prediction machineGenerative AINew content is made, which existed nowhere before. That is what generative means, and that is where the question of accuracy arises.
A photo application detects faces in order to set the focus.
Floor 1 · The Modela solid block on its own: the prediction machineAIThe useful counter-example: this is computer vision, so it is AI, and it has been there for a long time without anyone calling it that.
A service automatically sorts incoming email, then drafts a suggested reply.
BothThe sorting belongs to the first category, the drafting to the second. Most real systems combine the two, and they have to be evaluated separately.
The test that settles it
Ask what the system returns at the end. If it returns a decision, a score or a category, it is AI in the broad sense. If it returns a text, an image or a sound that did not exist, it is generative.
The trap
Saying “AI” while thinking of a conversational assistant makes the systems that decide in people’s place disappear from the debate. Yet those are the ones carrying the heaviest stakes, and they generate nothing.
Check click your answer
Level 1 · Recognise
Does a system that sorts job applications count as generative AI?
Level 2 · Distinguish
What relation ties artificial intelligence and generative AI together?
Level 2 · Distinguish
Why does confusing artificial intelligence in general with generative AI get in the way of writing an internal policy?