AI Research
The AI Neuron Myth May Be Cracking
The way researchers inspect AI could be due for a rethink—and the implications reach from model safety to the tools built around it. But a striking set of examples is not yet proof that neural networks are transparent.
Read article→