Current AI Technology and Transparency in Training Data

The discourse on current AI technology involves balancing between appreciating what is available now and envisioning future capabilities. It is crucial not to dismiss current AI technology due to unmet expectations. The recent EU AI Act emphasizes transparency in the training data of AI models but does not mandate revealing trade secrets. The quality and quantity of training data are pivotal for AI model success, as seen in cases where proprietary data sets differentiate between models. Analogous to the food industry's regulation on not disclosing all ingredients under trade secrets, AI models might use copyrighted data without it being part of the final model output, raising questions on transparency and proprietary information.