(Credit: Sabi via Wired)
At 130+ words per minute, I can type faster than most, but one Silicon Valley startup called Sabi is looking to make that all moot. Its engineers are working on non-invasive sensors that can detect brain waves from outside your head, and convert them into typed text, effectively allowing writers like me to write without typing at all, Wired reports.
In reality, this technology's benefits would extend far beyond providing mere typing speed parity. For those with injuries or disabilities that make communication difficult, a technology like this would be life-changing. The fact that the sensors would be embedded in a beanie or baseball cap, rather than in the brains of users, also gives Sabi a unique selling point over contemporary competitors like Neuralink.
"The biggest and baddest application of BCI [brain-computer interface] is if you can talk to your computer by thinking about it," Sabi investor Vinod Khosla tells Wired. "If you're going to have a billion people use BCI for access to their computers every day, it can't be invasive."
(Credit: Sabi)The invasive nature of BCIs like Neuralink is one of their main drawbacks. Sensor degradation, rejection, and the implant surgery make mainstreaming such technologies difficult, even if they might represent a more comprehensive solution to mobility issues for affected individuals.
Sabi's sensors can detect electroencephalography (EEG) from outside the head by massively scaling sensor density. Whereas traditional EEG sensors use tens or hundreds of sensors, Sabi uses tens of thousands, thereby improving the resolution of the data it records.
Translation is arguably the most complicated component in Sabi's proposed device. To convert brain patterns that can shift depending on mood, substance intake, fatigue, and a range of other factors, Sabi is leveraging a foundational AI model trained on a wide neural dataset to find common patterns in speech that enable thought-to-text conversion.
Fortunately for my TypeRacer.com records, Sabi is targeting an initial conversion speed of 30 words per minute, which is expected to improve over time, along with the accuracy of the text conversion. Combining ease of use with the ability to fit the sensors in beanies, caps, and other wearable headpieces could make Sabi an exciting opportunity for those with speech impairments. And eventually, perhaps, my slower-typing colleagues.