Generative AI models are trained on publicly available web data and further refined using licensed content. Tech companies are increasingly using public data on their own platforms to build distinct AI content libraries. However, once data is fed into a model, it is nearly impossible to remove or un-teach the AI. Companies like X (formerly Twitter), Microsoft, Instacart, Meta, and Zoom believe their data is best for training AI models. Elon Musk has threatened to sue Microsoft over the use of data from X, while Meta has excluded public LinkedIn data from its training set due to privacy concerns.
Key takeaways:
- Training data is a crucial factor in the competition among tech companies for generative AI. Large user networks create positive feedback loops that can be used to feed AI models.
- Microsoft CEO Satya Nadella has criticized other tech companies for locking up exclusive deals for content needed to train their AI models. Google's lead lawyer, John Schmidtlein, argued that Microsoft's data deficit is due to strategic errors and an inferior search product.
- Generative AI models are typically trained on publicly available web data and further refined using licensed content. Tech companies are increasingly using public data on their own products to build out their AI content libraries.
- Companies including X (formerly Twitter), Microsoft, Instacart, Meta, Zoom and more are updating their privacy policies to allow for user data to be collected to train AI models. Elon Musk has threatened to sue Microsoft over the use of data from X, while Meta has excluded public LinkedIn data from its training set due to privacy concerns.