This AI translates thoughts into text 🧠

A tool developed by researchers could one day allow people with aphasia to translate their thoughts into written text. This advancement in artificial intelligence (AI) opens up prospects for improving communication for patients with language disorders.


The problem of aphasia, which affects approximately 300,000 people each year in France, is well known: it is an impairment of the ability to speak and understand language, often following a stroke. This inability to communicate disrupts the daily lives of patients. Today, researchers at the University of Texas at Austin propose a major breakthrough by using AI to translate patients' thoughts into text, thereby facilitating their communication with those around them.

A tool to translate brain activity

This tool, developed by Alex Huth's team, is based on a brain decoder capable of translating brain activity into continuous text. Previously, such a system required 16 hours of training, during which the patient remained immobilized in a scanner while listening to podcasts. This constraint, which was difficult to endure for people with aphasia, has been reduced thanks to technological improvements.

The new version of the tool allows for training in just one hour. Thanks to an algorithm that adapts a patient's brain activity to that of another patient who served as a model for training the decoder, the process is now much faster and more accessible.

Thoughts beyond language

The researchers observed that the decoder does not directly translate language. In fact, the brain activity perceived during the reception of a story, whether oral or visual, is similar. This means that the decoder captures cognitive representations that are not necessarily related to language but to broader, universal concepts.

During this work, the researchers tested the decoder with participants by simulating lesions similar to those of aphasic patients. Although the tests were not conducted with real aphasic patients, the results are promising. The decoder successfully translated thoughts into text effectively.

A future for aphasic patients

The researchers are confident that this technology, after further adjustments, will be able to work on aphasic patients. Collaborations with aphasia experts are underway to test the tool in real-life situations.

The research team is also working on optimizing the interface to make it as accessible as possible. Their goal is to make the tool not only effective but also easy to use for people with language disorders.

This research represents a major step forward in the field of brain-machine interfaces, which promise to make communication much smoother for people suffering from aphasia. The reduction in training times and the improvement in results open up considerable prospects for the future.

To go further: what is aphasia and how does it affect communication?

Aphasia is a language disorder usually resulting from brain damage. It causes difficulties in speaking, understanding language, or writing, depending on the affected area of the brain. This disorder can be partial or total, making communication increasingly complex for patients.

The causes of aphasia are diverse, including strokes, head injuries, or neurodegenerative diseases. As a result, affected individuals may experience significant discomfort in their daily lives, often requiring therapeutic support to improve their communication skills.

Treatments for aphasia vary depending on the severity of the disorder and may include speech therapy. However, for many patients, full recovery remains difficult, highlighting the importance of technological assistance tools, such as AI, to improve their communication.