Sebastian Raschka: Learning ML, Responsible AI, AGI | Learning from Machine Learning #4
Learning from Machine Learning
00:00
The Dangers of Overconfidence in Machine Learning Models
The chapter discusses the lack of confidence levels in machine learning models and the dangers of overconfidence in neural networks. The speakers talk about the difficulties in obtaining and trusting confidence scores, using examples such as chat GPT and a classifier for AI-generated text. They mention ongoing research in addressing these issues and discuss alternative approaches for evaluating model outputs.
Transcript
Play full episode
