How AI knows what you’ll think next…

The Internet of Future Thoughts – Gautam Hazari
Imagine, sometime shortly in the future, on a relaxing Sunday, you pick up your smart glasses and put them on. The multi-channel EEG sensor-electrodes on the smart glasses detect your current state of mind and decide to project an augmented reality overlay of a short movie generated right at that moment by a generative video model based on the specific instructions from an action model deployed locally in the smart glasses microchip.
No, this is not a script from science fiction; it is feasible using today’s technologies.
Now, let’s imagine further. The short movie generated in real-time adapts to what your brain will be “thinking” in the next few seconds. It adapts the plot, the background score, and the location and even adds new characters.
And yes, all these may become a reality in the future.
A fascinating paper, “Predicting Human Brain States with Transformer,” was published in December 2024. [Read it here]. Yes, you read that right—it predicts and not just classifies the current brain state using a trained machine-learning model.

These researchers achieved a groundbreaking feat: predicting the brain state with fantastic accuracy—a mean squared error of 0.0013.
The transformer-based model, using resting-state fMRI (functional Magnetic Resonance Imaging) images, successfully predicted human brain states up to 5.04 seconds.
It’s like looking into the future of our thoughts in our brains.
I spoke and wrote about the IoTh, or “Internet of Thoughts” [Read here], last year. This is the Internet of Thoughts with a time machine – the Internet of Future Thoughts.

Let’s start with some basics of fMRI. The relative levels of oxyhaemoglobin and deoxyhaemoglobin (blood oxygen levels) in different brain regions indicate our brain states.
This can be detected using MRI scans, with an approach called BOLD (Blood Oxygen Level Dependent) contrast imaging. These BOLD “signals” are a language that explains what state our brain is in at that moment. This is the fMRI approach.
The researchers used large-scale and high-quality fMRI data from the Human Connectome Project (HCP), using the encoder-decoder structures of the transformer.
The outcome was 5.04 seconds of brain state predicted from the previous 21.6 seconds of brain states identified by the fMRI images!
This tells us what our brain will be thinking for the next 5.04 seconds in the future based on the previous 21.6 seconds. Incredible.
This is groundbreaking, challenging many fundamental aspects of our beliefs as a species and society.
What happened to our free will and agency? Was free choice an illusion? Is this adding more narratives to the “Box Universe Theory,” where time is an illusion and the past, present, and future all exist simultaneously in the “box” of the universe?
It indeed stirs up many philosophical paradigms.
However, considering the practical complications in our ever-converging organic and digital lives is even more intriguing. It can certainly be a lifesaver in deciphering many mental health issues, including critical mental disorders.
At the same time, it adds multiple new dimensions to our ever-changing, physics-defying, accelerating digital world:
Just imagine – the fraudster calling the innocent, unassuming human user using a deepfake video call, and already knows what the human user will be thinking in the next few seconds.
This is much beyond the GenAI induced Amygdala Hijacking, which I wrote about last year [Read here]

This is the next step to the “Internet of Thoughts” – it is the “Internet of Future Thoughts”!
Our future thoughts, the “thoughts” that will come to our brains in the future, are connected to the Internet. And it is raising many thoughts too.
Can this be a reality today?
Not really; it’s a little in the future.
Wearable MRI scanners are not a mass-scale reality, but this study gives directions to other brain scanning methods that may be used in the future for similar prediction models.
As with any technological evolution, humanisation must be at the core. It is in the interest of our species to be prepared and work together to find humanised solutions through technology.
At Sekura.id, we are busy creating and building the missing Identity Layer for the Internet.
Our vision is to make the digital world safer, and this vision extends to the “Internet of Things,” the “Internet of Thoughts,” and the “Internet of Future Thoughts.” The entire Sekura.id family is passionately working on humanised technological solutions.
Let’s make our digital world a SAFr place together.
Let’s work towards ID4All and Everything.
And yes, for every thought for now, and for the future, too.
Gautam Hazari is a relentless innovator with a knack for turning complex tech into seamless, human-centred solutions. As CTO of Sekura.id, he’s behind the mobile identity APIs that have redefined digital security, leading the rollout of GSMA’s Mobile Connect across 60 operators in 30 countries. His portfolio of identity and access control patents reads like a blueprint for the future of authentication.
A passionate advocate for ditching passwords forever, Gautam’s expertise stretches across AI, blockchain, and the Metaverse—topics he dissects with precision in articles, advisory roles, and TEDx talks. And then there’s his musical side: playing over a dozen instruments with the same finesse he brings to tech. Gautam Hazari isn’t just a CTO; he’s a pioneer crafting tomorrow’s identity landscape while finding time to soundtrack it.