Systems that produce rather than classify.
THE DISTINCTION
Traditional machine learning classifies or predicts: is this spam, what will this cost.
Generative AI produces new content: text, images, audio, code, video.
HOW IT WORKS BROADLY
The system learned patterns from very large amounts of existing content.
It generates new content following those patterns.
WHAT IT PRODUCES
- Text: articles, summaries, code, email
- Images: illustrations, photographs, designs
- Audio: speech, music
Video
WHY IT FEELS DIFFERENT
The output is novel rather than selected from a set. Each generation is new.
WHAT IT IS NOT DOING
Copying its training data, usually. It learned patterns, not passages.
Though models can reproduce memorised content, particularly text that appeared many times.
THE PRACTICAL IMPLICATION
It produces plausible content, which is not the same as correct content.
WHERE IT IS MOST USEFUL
First drafts, variations, and structured transformation of material you provide.
