Elon Musk’s habit of hyping a new technology, even when he is not a leader in the field, has long irked rivals.
It was on full display this week when he wrote on X that one of his companies, Neuralink, had succeeded in implanting an electrode into a human brain for the first time. The procedure is a small step towards Musk’s promise of being able to wire brains directly to computers to enhance their processing powers and one day enable humans to match the capabilities of advanced artificial intelligence.
Similar implants have been a staple of science labs for years. At least three rival start-ups have succeeded in inserting electrodes and using them to collect and interpret human brain signals.
But, while Musk’s claims bring a shrug of resignation from rivals, they also draw grudging acknowledgment that his attention grabbing has helped to propel the field closer to reality at a moment when it is reaching an important inflection point.
Musk has “really put a spotlight on this field and it’s bringing the [venture] capital in”, said Tom Oxley, chief executive of Synchron, which carried out its first human trials in 2019 and has raised $130mn.
Creating a storm of publicity about his efforts, even when there are few details to back up his claims, “is what Elon Musk does better than anyone else”, said Anne Vanhoestenberghe, professor of active implantable medical devices at King’s College London.
“Are they [Neuralink] ahead? No. Is their technology unique? No, none of what I have seen is novel,” Vanhoestenberghe said — though she credited Musk’s company with being “very advanced” and “state of the art” in the field.
Musk has long used his huge following on X and in the media as a strategic weapon to attract capital and talent to his ventures. It is an advantage he has exploited in his attempts to turn his new AI company, xAI, into a challenger to OpenAI, despite starting years behind.
Neuralink has raised nearly $700mn, much of it from Musk, and numbered some of the top brains in the field among its co-founders — though most of the founding team has since left the company, some to start rivals.
However, Musk’s ambitious claims also have a downside. They have led to “a huge weight of over expectations”, said Yurii Vlasov, a professor at the Grainger College of Engineering at the University of Illinois Urbana-Champaign.
Musk’s tweets have shone a light on a technology that strikes many as far-fetched, but which has recently started to return promising results.
Work on brain computer interfaces, known as BCI, began in earnest two decades ago but progressed slowly. After co-founding Neuralink in 2016, Musk promised human trials as early as 2020 but struggled to get US Food and Drug Administration approval.
Advances in materials and manufacturing have brought more refined, smaller arrays of electrodes, using techniques developed in semiconductor manufacturing. A second area of rapid development has been in the miniaturised electronics needed to amplify and channel brain signals. The final, and more recent, breakthroughs have come in the machine learning software needed to decipher brain signals, using them to drive a computer cursor or a prosthetic limb.
Neuralink’s method involves inserting extremely thin strands of electrodes into brain tissue with the aim of collecting electrical signals from individual, or small groups of, neurons. The procedure requires removing part of the skull to allow a robotic surgeon, which Musk has called a “sewing machine”, to thread the strands.
Neuralink has shown videos of monkeys using its implants to play Pong on a computer. However, the effects of having its electrodes sitting alongside brain tissue for long periods of time are not yet clear.
Other companies and academics have run human trials with other techniques and notched significant achievements in interpreting brain signals, said Thomas Hartung, a professor at Johns Hopkins University in the US who leads an international biocomputing project. He cited a team at Stanford University who reported in 2021 how they had converted a paralysed man’s brain signals from imagining handwriting into text on a computer.
Other techniques involve trade-offs between the invasiveness — and risks — of the implants and the quality of the signal collected.
Precision Neuroscience, co-founded by Benjamin Rapoport, a neurosurgeon and one of the founders of Neuralink, makes tiny slits in the skull to insert a mesh of microelectrodes that “wrap” around the brain. Though collecting less data than Neuralink’s electrodes, this less-invasive technique should still yield enough data to control a prosthetic limb, according to Rapoport. “It is not individual neurons that control muscle movements,” he added.
Synchron, meanwhile, inserts its sensors in the skull through a vein, in much the way a coronary stent is implanted — a method that it hopes will allow brain implants to become a routine procedure.
The brain signal it collects, though less detailed, should still be strong enough to achieve what Oxley called “product/market fit” — the goal, pursued by all tech start-ups, where a technology is good enough to produce a useful product. Synchron aims to channel brain signals to control a smartphone or a tablet, giving patients with partial paralysis more ways to communicate and control their environment.
These different techniques could lead to a range of products, said Alex Morgan, a partner at Khosla Ventures, which has invested in neurotechnology, including Synchron. “This isn’t a winner-takes-all” technology, he said.
Despite significant scientific breakthroughs, however, few in the field are willing to predict when the technology will bring useful products. A particular challenge has been to interpret the signals collected from the brain, making it difficult to tell when the technology will be able to do more than move a computer cursor or activate simple prosthetic limb movements, Vlasov said.
In consultation with the FDA, start-ups working on brain computer interfaces have converged on a similar goal: to develop implants for patients with the most severe forms of paralysis, Rapoport said.
That is a long way from the kind of mind-enhancing technology dreamt of by Musk. But, while Rapoport said that lies far in the future, he added: “I don’t think it’s inconceivable.”
Read the full article here