Skip to main content

Why Deep-Learning AIs Are so Easy to Fool

Douglas Heaven Nature
Brilliant at what they do, taken into unfamiliar terrain Deep Neural Networks are fundamentally brittle. AI researchers say they are making progress in fixing the flaws, but acknowledge that they’re still groping for new techniques.

New Theory Cracks Open the Black Box of Deep Learning

Natalie Wolchover Quanta Magazine
cartoon dog with its neck caught in a bottle A new idea called the “information bottleneck” is helping to explain the puzzling success of today’s artificial-intelligence algorithms — and might also explain how human brains learn.
Subscribe to deep learning