ALBAWABA - In a groundbreaking development, scientists have unveiled an artificial intelligence system capable of translating human thoughts into written words. The revolutionary technology utilizes a helmet equipped with sensors and artificial intelligence to achieve this remarkable feat.
As reported by NewScientist, a team of researchers demonstrated their ability to convert a person's thoughts into written text by having participants wear a sensor-laden cap, recording electrical brain activity as they read text passages.
The recorded electroencephalogram (EEG) data is then processed through an artificial intelligence model named DeWave. This model is trained by correlating specific brain signals with sentences, leveraging numerous examples for comprehensive learning.
For instance, when a person thinks of saying 'hello,' distinct brain signals are sent. DeWave learns the association between these signals and the word 'hello' by analyzing a variety of examples for different words and sentences.
Once DeWave comprehensively understands the nuances of brain signals, the team integrates it with an open-source large language model (LLM), akin to the technology powering ChatGPT. The LLM is instructed to function as an intelligent writer, utilizing DeWave's signals as a guide to construct coherent sentences.
In the final stage, the researchers jointly trained DeWave and the language model to further enhance their collaborative capabilities, resulting in improved sentence generation based on EEG data.
Scientists anticipate that as this system undergoes further refinement, it holds the potential to significantly aid communication for individuals who have lost the ability to speak, such as stroke survivors, marking a breakthrough in patient care.
