This year gave rise to an incredible mix of brain implants that can record, decode, and alter brain activity.

It sounds like déjà vu—brain-machine interfaces also lived rent free in my head in last year’s roundup, but for good reason. Neuroscientists are building increasingly sophisticated and flexible electronic chips that seamlessly integrate machine intelligence with our brains and spinal cords at record-breaking speed. What was previously science fiction—for example, helping paralyzed people regain their ability to walk, swim, and kayak—is now reality.

This year, brain implants further transformed people’s lives. The not-so-secret sauce? AI.

One implant in the spinal cord of a patient with Parkinson’s disease—which slowly destroys a type of brain cell for planning movements—translated his intention to move. After decades, the man could once again stroll down a beachside road with ease. The study paves the way for the restoration of movement in other brain disorders—like Lou Gehrig’s disease, where neural connections to muscles slowly disintegrate, or in people with brain damage from stroke.

Another trial used electrical stimulation to boost short-term memory in people living with traumatic brain injuries. The carefully timed zaps increased attention span decades after the injury—allowing participants to juggle multiple everyday tasks and pursue hobbies like reading.

Brain implants also thrived as diagnostic tools. One study used implants to decode brain wave patterns associated with depression and to potentially predict relapse. The study deciphered how brain signals differ between a healthy and depressed brain, which could inspire better algorithms to nudge brain activity away from depression.

But perhaps the greatest progress was in decoding speech—technologies that translate thoughts into words and sentences. These technologies support people who have lost the ability to speak, giving them an alternative way to communicate with loved ones.

Here are the 2023 highlights from a new generation of “brain-reading” implants.

Thoughts to Text

We speak at a rate of roughly 150 words a minute. It’s a high bar for brain implants.

Many neurological disorders, such as stroke, paralysis, or locked-in syndrome, rob a person of the ability to speak—even if their mind is still coherent. Early this year, a Stanford team helped a 67-year-old woman restore her speech at 62 words a minute, over three times the speed of previous implants. The woman lost her voice due to Lou Gehrig’s disease, which slowly erodes the brain’s ability to control muscles for speech, movement, and eventually breathing.

The study used a massive library of words to decode her speech from two sources: electrical activity in Broca’s area, the brain’s “language center,” and from muscles around her mouth. These signals were fed into a recurrent neural network—a type of deep learning algorithm—to distinguish the basic elements of speech. In just three days, the system was able to decode the woman’s thoughts at record speed—although with errors.

Another system went one better. Rather than using electrodes that penetrate the brain, the device—called ECoG for electrocorticography—consists of small plate-like electrodes placed on the surface of the brain to capture electrical signals. It still needs to be implanted beneath the skull but limits damage to the brain’s sensitive tissues. Each electrode, roughly the size of the head of a thumbtack, can record high-quality neural signals.

ECoG was first used at the turn of this century to record speech and movement signals in people with epilepsy. It soon developed into a device that allowed a person with locked-in syndrome to communicate their thoughts using the implant at home.

What’s new is the introduction of AI. Some algorithms decoded the brain activity of vocal movements—for example, the position of the tongue and shape of the mouth—while large language models, like those powering ChatGPT, constructed sentences from the data. Though the system could translate brain signals to text at roughly 78 words a minute, roughly a quarter had errors. But non-verbal communication made up for the mistakes: the implant used facial expressions to animate a digital avatar, giving patients yet another mode of communication.

A Turning Point

Brain implants are a type of brain-machine interface. True to their name, these devices link the brain to computers. How they bridge the two is wide open to creative solutions.

Most systems measure electrical activity in the brain and often require cables that link electrodes to computers that can decode neural activity.

This year, a study cut the cord with a wireless implant. The system consists of flexible, grain-sized circuit boards sprinkled across the brain that can detect and temporarily store changes in activity. These “nodes” wirelessly transmit data to a headphones-shaped receiver, which processes the information, controls brain stimulation via the nodes, and powers the array. Although wireless, the system still requires surgery for implantation.

An alternative? Devices that capture brain signals without surgery.

One study used AI to translate functional magnetic resonance imaging (fMRI) data—a non-invasive technique—into the “gist” of a person’s thoughts. The technology doesn’t translate brain activity into words; instead, it captures ideas as they evolve, even though the exact words are lost in translation. Another study measured brain activity with swimming-cap-like headgear embedded with electrodes that sits on the scalp. As a user silently read sentences in his mind, the cap—with the help of AI—translated his “thoughts” into text.

Other devices are exploring entirely new methods of connecting machine to brain—for example, with light. One recent study combined neurons genetically engineered to respond to light and flexible probes that activate these neurons with different colors of LED light. Combined with a common technology that controls light settings, the device, with over a thousand independent LED pixels, could control the activity of multiple individual neurons at once.

Brain cells are noisy. The new device helped sort through the cacophony to resolve the brain circuits underlying specific mental roles. It activated neurons up to five millimeters deep inside a mouse brain—roughly the thickest part of the human cortex.

Paradigm Shift

Brain implants aren’t mind-reading machines. But as the technology evolves, it’s likely to encounter numerous ethical landmines.  A device broadcasting thoughts as text, for example, could inadvertently encroach on privacy.

The United Nations Educational, Scientific, and Cultural Organization (UNESCO) is already looking ahead. This summer, they released a blueprint on neurotechnology, calling for global regulations and an ethical framework as brain implants hurtle towards an unknown future. The organization previously developed similar guidelines for other key breakthroughs, such as how to use and share human genetic data and how to develop AI to improve society as a whole.

Brain implants have been moving fast, but their real-world utility is just getting started. With transformative power comes responsibility. A global conversation on access, equality, privacy, and more philosophically, what it means to be human shouldn’t be an afterthought. Rather, it may be as important as the tech itself as we continue into an era of cyborgs.