This talk summary is about Tamuz Hod, a Signal Processing and ML Infrastructure expert, discussing the use of machine learning techniques to improve small manipulation and increase alignment in AI systems. Hod highlights the major bottlenecks in small manipulation, such as image labeling and the need for skilled human labor, which can be alleviated using machine learning. They also emphasize the importance of increasing the bandwidth for human feedback and understanding human values in AI algorithms. To achieve this, Hod explores the potential of Brain-Computer Interfaces (BCI) and personal capability development. They invite others to share their thoughts and experiences on using BCI for alignment research and emphasize the need for higher bandwidth feedback to improve the effectiveness of AI systems.