University of Texas at Austin researchers have created an artificial intelligence (AI) system capable of reading and reproducing human ideas.
The researchers recently released a paper in Nature Neuroscience in which they investigated the use of AI to non-invasively convert human thoughts into words in real time.
Current approaches for decoding thoughts into words, according to the researchers, are either intrusive — requiring surgical implantation — or limiting in that they “can only identify stimuli from among a small set of words or phrases.”
Read Also: ChatGPT, Midjourney, other AI tools to make their way into EU legislation
The Austin team worked around these constraints by training a neural network to decode functional magnetic resonance imaging (fMRI) information from several areas of the human brain at the same time.
The researchers had multiple test volunteers listen to hours of podcasts while an fMRI equipment monitored their brain activity non-invasively. The generated data was then utilised to train the algorithm on the mental patterns of a single user.
Read Also: 7 popular tools and frameworks for developing AI applications
Following the training, test subjects’ brain activity was monitored again while they listened to podcasts, watched short films, and silently imagined telling a story. The AI system was supplied the subjects’ fMRI data during this part of the trial and decoded the signals into plain language in real time.
According to a news statement from the University of Texas at Austin, the AI was correct around half of the time.
Fortunately for anyone concerned about AI infiltrating their thoughts against their will, scientists have stated unequivocally that this is not currently a possibility.
The system can only work if it has been trained on a specific user’s brainwaves. This renders it ineffective for scanning people who haven’t spent hours supplying fMRI data. Even if such data was generated without a user’s permission, the team eventually finds that both data decoding and the machine’s ability to monitor thoughts in real time require active participation on the part of the person being scanned.
However, the researchers did note that this might not always be the case:
“Our privacy analysis suggests that subject cooperation is currently required both to train and use the decoder. However, future developments might enable decoders to bypass these requirements. Moreover, even if decoder predictions are inaccurate without subject cooperation, they could be intentionally misinterpreted for malicious purposes.”
In similar developments, a group of Saudi academics recently discovered a way for enhancing precision in brain tumour diagnosis by processing MRI scans through a blockchain-based neural network.
The Saudi researchers show how processing cancer research on a secure, decentralised blockchain might improve precision and eliminate human error in their article.
Read Also: Elon Musk to launch truth-seeking artificial intelligence platform TruthGPT
While both of the aforementioned experiments are cited as early work in their respective research papers, it is important to note that the technology used in each is widely available.
The AI at the heart of the team’s work at the University of Texas in Austin is a generative pre-trained transformer (GPT), the same technology that ChatGPT, Bard, and other large language models are built on.
Furthermore, the Saudi Arabian team’s cancer research was carried out using AI that had been trained on Nvidia GTX 1080 GPUs, which have been available since 2016.
In reality, nothing prevents a bright developer (with access to an fMRI machine) from merging the two ideas to create an AI system that can read a person’s thoughts and record them to the blockchain.
This might pave the way for a “proof-of-thought” paradigm, in which people could mint nonfungible tokens (NFTs) of their thoughts or keep immutable ledgers of their sentiments and ideas for posterity, legal purposes, or simply bragging rights.
The significance of mind-to-blockchain NFT minting, for example, may have ramifications for copywriting and patent applications in which the blockchain serves as proof of when a thought or idea was recorded. It may also enable celebrity intellectuals, such as Nobel laureates or current philosophers, to codify their thoughts in an indelible record that can be commoditized and used to create collectible digital goods.
