Artificial intelligence can read our thoughts
Many neurotechnology companies are developing brain sensors capable of decoding our state of focus, our emotions and even the content of our mind.
The development of artificial intellligence continues its frantic race and will even accelerate further. Recent massive investment announcements in this industry attest to that. In the United States, a new venture called Stargate uniting OpenAI, Oracle and Softbank will spend 500 billion dollars in the coming years to build data centers dedicated to AI.
In France, a few days after the launch of the Artificial Intelligence Action Summit which took place in Paris from February 6 to 11, president Emmanuel Macron said on France Televisions that investment funds and private companies will invest 109 billion euros to make France an AI powerhouse. Shorty after, the president of the European Commission Ursula von der Leyen announced that the European Union wants to mobilize 200 billion euros for AI investments in Europe. Why spend such stratospheric sums on AI? What are the objectives of these technology companies and the governments which they are connected to?
For many, artificial intelligence is the generative AI that we find in chatbots, these conversational agents like ChatGPT from OpenAI, Gemini from Google, Grok from xAI or Deepseek from the chinese company of the same name. Generative AI programs to create images and videos have also been quickly adopted by the general public.
We also hear more and more about health applications of AI, for cancer detection as an example. When he was asked by American president Donald Trump about the uses of the AI systems which will be trained thanks to the Stargate project, co-founder and Chief Technology Officer of Oracle Larry Ellison only spoke of the health sector, promoting Messenger RNA anti- cancer vaccines made with AI. But is it really to improve people’s health that these technology companies spend so much money developing AI systems? We have every right to doubt the allegedly altruistic impulses of these hypercapitalistic corporations.
In this video, Larry Ellison of Oracle explains that AI surveillance will keep citizens “on their best behavior” because “we’re constantly recording and reporting everything that’s going on”.
What if one of the answers to Trump’s question was actually to be found... in our head?
Parallel to the development of generative AI, the most visible aspect of artificial intelligence and the one on which most public debates focus on, many technology companies are moving ahead in other sectors. Often discreetely, away from scrutiny and democratic conversations. Weapons and surveillance firms are amongst these. But neurotechnology companies also benefit from the advances in artificial intelligence.
Neurotechnologies
Neurotechnologies comprise an ensemble of methods and electronic devices which interact with the nervous system to analyze, record and modify brain activity.
Private investments in neurotechnology companies have seen a 22-fold increase (+2200%) between 2010 and 2020, reaching a total of 33.2 billion dollars in 2020, according to a Unesco report published in July 2023. Governments are also very active in this space, with more than 6 billion dollars invested from 2013 to 2020. In 2025, the global neurotechnology market size is estimated at 17.3 billion dollars and should go beyond 50 billion dollars by 2034.
One of the most famous neurotechnology companies is Neuralink, co-founded in 2016 by South-African businessman Elon Musk, the richest man on the planet. On January 29, 2024, he announced that his company has implanted a chip in a patient’s brain for the first time. Shortly after, Musk wrote on X that the first product of his company is called Telepathy and that it will enable people to control their phone and computer, and thus many other devices, just by thinking. Another product from Neuralink is called Blindsight and will enable blind persons to see again.
Therapeutic applications of neurotechnologies promise to be vast: enable paralyzed people to walk again, reduce or eliminate epilepsy crisis, command a prosthetic or a wheelchair by thoughts, as well as detect and treat psychiatric pathologies such as compulsive obsessive disorders, schizophrenia, Alzheimer and Parkinson’s disease.
It goes even further: neurotechnologies can read our thoughts.
AI analyzes the brain
Artificial intelligence accelerates the development of neurotechnologies thanks to its capacity to analyze vast quantities of brain data. (Read the United Nations report “Impact, opportunities and challenges of neurotechnology with regard to the promotion and protection of all human rights”, p.3) Generative AI in particular is now capable of decoding neural informations that were previously indecipherable, such as the words someone is thinking.
In a study published in December 2023, researchers at the University of Technology Sydney have connected an electroencephalogram cap (which measures the brain electrical activity) to large language models (deep neural networks such as those used by the chatbot ChatGPT). Participants were reading phrases silently as an AI called DeWave was analyzing their brain data to translate it to words.
In this video, we see a man equipped with an electroencephalogram (EEG) cap who thinks for example of the sentence “Yes, I’d like a bowl of chicken soup, please”. The DeWave AI then translates his brain data as “Yes, a bowl of beef soup.”
This neurotechnology “brain-to-text” reached a precision score of around 40% in this study, one of the first of its kind. Researchers at the University of Technology Sydney think they will achieve a precision of 90% in the future, which means this AI will be able to read almost all thoughts down to the last word...
Mind surveillance
The DeWave AI is even more impressive in that it is used with a non-invasive device, a EEG cap, which doesn’t need to be placed inside the skull. This is not the case of other “brain-to-text” technologies which require the insertion of a chip inside the brain, or at the surface of it, with a chirurgical operation. Furthermore, a EEG cap is a light device, portable and affordable, unlike a magnetic resonance imaging machine, which also allows for the decoding of thoughts.
Many companies throughout the world already use non-invasive brain sensing devices to monitor their employees. These sensors can’t read their thoughts yet because the “brain-to-text” technology is still under development, however they can detect if their employees are not focused enough.
For the past decade, Australian company SmartCap Technologies have sold its neurotechnology devices to over 5000 companies worldwide active in the mining, construction, trucking, aviation and railway industries. Their purpose? To make sure their employees are always wide awake, thanks to a brain sensor placed inside their hard hat or their cap which sends a signal when they become drowsy. (Read The Battle for Your Brain by Nita Farahany, pp.40-41)
In the United States, the company Emotiv created the MN8, a pair of earbuds equipped with brain sensors. With only two electrodes, one in each ear, the MN8 device allows the employers to monitor the emotional and cognitive functions of their employees in real time. These earbuds also send a signal to the worker and her boss when they detect a lower state of focus. (Read The Battle for Your Brain by Nita Farahany, p.48)
In 2019, a chinese school in Jinhua used brain sensors on their pupils to track their concentration levels. This pilot project was then cancelled due to public outcry. It is likely, however, that other schools will also use similar neurotechnology devices later on.
Brain sensors are thus already adopted by thousands of companies and will be used more and more in the future, whether at work or at school. Consumer grade devices are also available, for example to facilitate meditation, improve sleep quality or play video games. It is only a matter of time before these objects will also be capable to read the thoughts of their users.
A report on brain sensors from Context. The video starts by showing that it is already possible to order some models on the internet.
A new gold rush: our brain data
Artificial intelligence needs lots of data to be trained on. Given that almost all the publicly available internet data has already been analyzed by big AI companies, they now need new data to keep improving their models. In this context, neurotechnologies represent an attractive opportunity for these firms because they allow them to collect massive amounts of brain data, which are very rich in informations. Many companies will seize this chance and therefore mind surveillance will become more and more widespread and intrusive.
This creates a lot of risks, even more so because these neural sensors can also modify brain activity: mass surveillance, complete loss of private life, manipulation of people’s thoughts, memories, emotions and body, brain data thefts, cyberattacks... Aware of these dangers, neuroscientists and institutions such as Unesco are mobilizing to protect our neurorights, especially our right to mental privacy, emphasizing the urgency to act because of the very fast development of neurotechnologies. This will be the topic of my second article.
– Arnaud Mittempergher
Thank you for reading the very first article of Artificial reality! If you liked it, you can subscribe for free to receive a press review every week on the latest important news in artificial intelligence. With a paid subscription, you will also have access to all articles and to an exclusive monthly newsletter.
