Brain interfaces should help AI agents understand what people want, not merely substitute for keyboards and mouse clicks, Rahul Chhabra argues in “What if AI could read your mind?”, which he shared on Oct. 17. He proposes using decoded intentions as prompts, with brain signals also helping personalize AI around taste and preferences.
Chhabra describes a baseball cap that Sabi is building and claims the company has trained a typing brain-computer interface model that can help predict a user’s next three to four keystrokes. He presents agents taking anticipatory actions in digital and physical spaces as a prospect if such models scale to more open domains.
Brian Roemmele endorsed the argument, writing: “typing was never the point. AI that understands what you mean is.” He also invoked Hans Berger, whom he described as inventing EEG in 1924 with telepathy as his goal, and speculated that AI might bring that ambition closer.
Chhabra makes trust a condition: neural data must remain encrypted, he argues, and nobody other than the user should have access to what they are thinking.