This is how AI Midjourney imagines a mind-reading AI.Photo: Watson/Midjourney
Follow me
Reducing the achievements of two University of Austin scholars to a few words, one might address:
“Artificial intelligence can read minds.”
Five words that make you uncomfortable. But it is not (yet) that simple.
In fact, Jerry Tang and Alex Huth, a computer science graduate student and assistant professor of computer science and neuroscience, were able to convert the brain activity of various subjects into texts. An interview paper was published in early May “normal neuroscience” published. However, the system only works under certain conditions. This is very broad:
- Similar to the Llms (Large Language Models) of well-known AI systems, this system must first be “trained”. For this purpose, subjects’ brain waves are recorded while listening to radio plays. This training takes hours for each individual.
- The system is only 50 percent correct.
- Interpretation is only successful if the person tested is cooperative and “tells” a story in their head, listens to a radio play or watches a movie.
On the other hand, correct interpretations cannot be achieved with people who …
- …which AI is not trained in.
- … Actively resist trying or try to think of something else.
So the reassuring news is: no artificial intelligence can absorb our thoughts yet. No harm can be done to this technology so far: “We are well aware of the dangers of misuse,” Tang told Austin University Media Service. “We want to make sure that this type of technology only works with the consent of the people involved.”
The groundbreaking thing about the new model from the University of Texas is that brain wave receptors do not have to be surgically implanted into a subject’s head beforehand. Longer texts can be captured – not just individual words or sentences.
There are a few possible application areas. For example a “mouthpiece” for people with locked-in syndrome. Caged individuals are mentally healthy but unable to speak or move due to physical weakness. According to the study authors, there is still a long way to go before it can be used in real life.
Racism scandal: contractor slapping employees
Video: Watson/Aya Baalbaki
Spike, the sea lion, can play – that’s what’s behind it
Video: Watson
This may also interest you:
“Typical entrepreneur. Lifelong beer expert. Hipster-friendly internet buff. Analyst. Social media enthusiast.”