Creating an Immutable ‘Family Tree’ for AI Training Data

The race for every company to embrace AI has them combining myriad and often inconsistently documented datasets in their training, raising legal and ethical questions about where the data actually originated.

MIT’s Data Provenance Initiative audit of more than 1,800 datasets found that companies can’t reliably track where their AI training data comes from. In fact, the researchers have concluded that authenticity, as well as consent practices and provenance, are broken in AI, leaving companies without the training data transparency…