Scientists at the University of Texas in Austin have undertaken the first successful
experiment in AI-enabled mind-reading. Typically, about 90% of the brain’s
conscious activity is self-reflective – it can best be described as our inner dialogue.
Only 10% of our thoughts are directed outwardly and communicated. Identical twins,
lovers or people who are extremely close are frequently reported to be able to read
their counterpart’s thoughts.
They can, e.g., complete a sentence their other halves
had begun, or feel what they are feeling. As I explain in my book ‘Subliminal
Leadership’, mind reading and empathy are two, closely related manifestations of
‘inter-brain connectivity’, the ‘over the air’ synchronization of brain waves between
people. Inter-brain connectivity has been investigated in many scientific experiments,
but the underlying physical principles of information transmission are not yet
understood.
The experiment performed at the University of Texas achieved the complex task of
not only reading and interpreting brain wave frequency bands typically associated
with certain general brain states such as sleep or solving complex tasks, but specific
thoughts. The study conducted by Jessy Tang and colleagues had volunteers watch
silent movie clips from the ‘Sintel’ while their brain activity was recorded. Importantly,
the researchers applied non-invasive technology (instead of inserting electrodes into
the brain, they used a non-invasive magnetic imaging device to track brain
activity). From the recording, the AI was able to accurately describe what the
participants had been watching, essentially ‘reading’ their thoughts.
AI has already brought immense progress in fields such as medical diagnostics, drug
development, logistics and many other industries and the researchers pointed at the
advantages of neurotechnology-enabled mind reading for speech-deprived patients
such as in the aftermath of a stroke. Yet, the world’s autocracies have already taken
notice and will have their own ideas about how this technology could be exploited,
no matter any future international regulation of AI.
Tang, J., LeBel, A., Jain, S. et al., Semantic reconstruction of continuous language from
non-invasive brain recordings, NatureNeuroscience Vol. 26, 2023, pg. 858–866