Imagine typing with the power of your mind. It sounds like science fiction, but new research on brain-computer interfaces (BCIs) is bringing this closer to reality. Researchers have developed ChatBCI, a system that uses the brain's electrical signals, combined with the predictive power of large language models (LLMs) like those behind ChatGPT, to let users compose sentences by simply thinking about what they want to say. BCIs typically rely on users painstakingly selecting letters one by one on a virtual keyboard by focusing their attention. This process is slow and mentally taxing. ChatBCI tackles this challenge by using a standard P300 speller, which detects a specific brainwave pattern when a user focuses on a desired character. However, instead of just letters, ChatBCI displays predicted words generated by a LLM. This LLM receives the initial letters typed by the user (through the BCI) and suggests possible words, dramatically reducing the number of selections needed. The user can simply focus on the correct word, and the system does the rest. In experiments, ChatBCI allowed users to compose sentences much faster than traditional BCIs, achieving a remarkable increase in typing speed. In one task, participants used ChatBCI to improvise sentences, showcasing its potential for natural, real-time communication. The study introduced a new way to measure the efficiency of predictive typing, called keystroke savings, demonstrating how LLMs can minimize user effort. This breakthrough opens up exciting possibilities for people with communication and motor disabilities. ChatBCI offers a glimpse into a future where thought-to-text technology becomes a practical, everyday reality. While challenges remain in improving accuracy and expanding the vocabulary, the integration of AI with BCIs marks a significant leap forward in assistive technology.