A Visual Intro to Machine Learning
Stephanie Yee and Tony Chu have published a clear, interactive introduction to machine learning that opens with a crisp definition: in machine learning, computers apply statistical learning techniques to automatically identify patterns in data, and those techniques can be used to make highly accurate predictions.
The piece is a two-column "scrollytelling" site that walks through how classification builds into a decision tree. It then shows what happens when new data is introduced and predictions become less reliable due to overfitting — a concept the authors signal they will tackle next. That follow-up will probably cover correcting the model, which matters for accuracy but also introduces a mechanism for fixing mistakes. That distinction is important, given how Weapons of Math Destruction repeatedly made the point that algorithms without a corrective model can be incredibly dangerous.
It is odd, the authors note, that machine learning tends to elicit a "cool! good job computers! let's use this!" reaction while artificial intelligence often gets a dismissive "meh, it's all just programmed algorithms" shrug — even though the underlying ideas overlap heavily. The footnote adds that some ML applications don't need much accuracy at all; the Wombo art generator, for instance, produces paintings from prompts presumably via ML, and while the results are fun, anything you make with it is owned by the company.
