DeepCast LogoDeepCast Full Wordmark

Topic: AI training data

AI training data refers to the datasets used to train machine learning models, which can embed human biases and have significant implications for the performance and fairness of AI systems.

More on: AI training data

The podcast episodes explore various aspects of AI training data, including how it can embed human biases and limitations into AI systems, the ethical and legal considerations around acquiring and using training data, and the importance of transparency around training data practices.

For example, episode The TED AI Show: Why we can't fix bias with more AI w/ Patrick Lin discusses how the training data used for AI models can propagate societal biases, while episode Firefly's Training Secrets, Falcon 9's 20th Flight, and the Podcast Ad Illusion covers the controversy around Adobe's use of competitor AI-generated images in training its own AI model, Firefly.

The episodes also touch on the need for high-quality, representative training data, as seen in Luis von Ahn Explains How Computers and Humans Learn From Each Other, which discusses Duolingo's use of CAPTCHAs and user data to enhance its language learning AI.

All Episodes