What to Know

• In a very real way, AI can read your mind by interpreting signals in your brain and extrapolating your thoughts from that information.
• It's a non-invasive process, which means the AI works by reading brain scans instead of relying on surgically-implanted electrodes.
• It leans on the same type of technology that powers ChatGPT.

How Artificial Intelligence Can Read Minds

Artificial intelligence reads minds by interpreting brain signals. Early attempts at reading minds in this way used invasive methods to measure electric impulses in the brain, and some were successful in coming up with single words or even short sentences.

Recent instances have leveraged deep learning algorithms and large language models to interpret non-invasive brain scans with significantly more robust results.

AI mind reading starts with a functional magnetic resonance imaging scan (fMRI) of a person's brain. Data from the scan is fed to a specially trained AI program that interprets it using a large language model in a process that's essentially the same as providing a text prompt to ChatGPT.

The AI analyzes the data and then determines what the person was thinking about based on its large database of other scan results.

Some experiments with mind-reading AI have turned peoples' thoughts, or even dreams, into images using stable diffusion and similar technologies. Other experiments have turned a person's thoughts into text reflecting what they were thinking about. 

How Is AI Trained to Read Minds?

Artificial intelligence is trained to read minds using deep learning algorithms and large language models. Instead of providing vast amounts of text, which is how a model like ChatGPT is trained, the AI is provided with brain scans and corresponding data to show what the person was thinking about at the time of the scan.

One of the early successes in mind-reading AI used a custom-trained version of GPT-1, an earlier version of the technology that powers ChatGPT.

In that instance, volunteers lay in MRI machines for hours while listening to podcasts to create data for the training process. The researchers then trained GPT-1 using the brain scans and corresponding podcast audio. Newer GPT models have built upon this approach.

Once an AI has been trained on brain scans and corresponding data, it can use that training to interpret other brain scans.

Mind-Reading AI for People with Paralysis and Other Conditions

One of the most important uses for mind-reading AI is to assist people with paralysis and other conditions. While there has been some success with using technology like Bluetooth to bypass injured spinal columns and to control robotic limbs, they have relied on invasive procedures that tap directly into nerves to read their signals.

Non-invasive mind-reading AI could control prostheses and other devices via brain scans instead of surgically-installed electrodes. This could allow someone with a missing arm or leg to control a prosthesis or a paralyzed person to communicate their thoughts via text or speech synthesis.

Artificial intelligence can mimic voices realistically, so someone who was paralyzed in an accident could theoretically even speak using their own voice if recordings were available to train the AI.

Is It Okay to Do This? A Look at the Ethical Considerations and Privacy Concerns of Mind-Reading AI

While there are a lot of exciting possibilities associated with mind-reading AI, there are also a number of ethical considerations and privacy concerns that come with the potential uses of this technology. Problems surrounding consent, privacy, and data security are all key issues here.

The issue of consent is especially problematic because the ethical implementation of this technology would require informed consent that could be difficult or impossible to obtain in some cases.

For example, this technology could allow a paralyzed person to speak, but obtaining the initial consent required to use the technology on such a person could prove difficult.

In other situations, such as an employer asking their employees to use the technology, the inherent power imbalance could also raise issues surrounding consent.

Privacy and data security are also large areas of concern in regard to mind-reading AI. Since this technology reads your mind, it can render the inner workings of your mind, and even your innermost thoughts, no longer private.

That is a consideration that potential users would need to grapple with, but the issue of losing that level of privacy also raises large concerns about data security.

Personal data obtained through data mining and other methods is already a major commodity, so it's likely that data obtained via this type of mind-reading technology would also be valuable and sought after.

Legislation might require additional layers of security for this intensely private information, and any data breaches could have extreme consequences.

• Where can I use AI?

  You are using a form of AI on your smartphones now when sorting photos and looking for common items like trees, beaches, people, etc. If you want to try it, you can use the Microsoft's Bing AI by signing up for an account.

• How do smart assistants work?

  This software works with hardware to interpret your spoken command(s) and answer the question or request. They are programmed with the most common uses ("Hey Siri, add cheese to the shopping list," for example) and, if they can't interpret the request, send the audio file to more powerful computers to interpret. Once it knows your request, it consults databases and tells you the information.