A former employee of Google's DeepMind has expressed concerns regarding the current state of large language models (LLMs), stating that the primary issue is a lack of sufficient training data. He noted that the Internet has been largely exhausted as a data source, leading to reliance on synthetic data for training. This sentiment is echoed by various industry experts who emphasize the importance of high-quality data in enhancing AI capabilities. One expert pointed out that the effectiveness of AI is contingent upon the selection of data that powers it. Furthermore, there is a growing movement towards user-owned AI, which aims to incentivize users for contributing their personal data to improve AI models. The current challenge is not merely acquiring more data but unlocking high-quality, user-owned data that has previously been inaccessible to AI systems. Platforms like Twin Protocol are highlighted for their role in empowering individuals to contribute valuable data, which is seen as crucial for the future of AI training.