Elon Musk gave the world a progress update on his brain-machine interface startup Neuralink on Friday, showcasing a small implant that can read and transmit the neural activity of a pig.

Why it matters: The Neuralink implant still has yet to be tested in human beings, but it's part of a wave of brain-machine interface technologies that aim to address neurological diseases and injuries, and eventually directly link human brains to the internet.

What's happening: In an online event, Musk showed the Neuralink device — roughly the size of a quarter — implanted in a pig's brain, where it was able to read some neural signals in real time.

• While Musk had billed the event as a "product demo," Neuralink has yet to undergo clinical trials in human beings, though the company announced it had been granted FDA "breakthrough device status," which could speed efforts in humans.
• "It's like a Fitbit in your skull with tiny wires," said Musk. The Neuralink's wires connect to brain centers, while the device communicates wirelessly to a computer.

Our thought bubble: While Musk said in 2019 that he would like to conduct clinical trials as early as this year, it's not clear when such experiments will actually occur.

Background: Work to use machines to read brain activity dates back to 1929, when Hans Berger recorded the first human electroencephalogram (EEG), which detects electrical activity in the brain.

• Today some version of neural implants are already used by tens of thousands of people around the world, mostly to treat neurological disorders like epilepsy and Parkinson's.

The next generation of brain-computer interfaces (BCIs) seek to translate brain signals into useful data or even use them to manipulate machines.

• In 2017 Facebook announced that it wanted to create a headband that would allow people to type with just their thoughts, and last year research funded by the company showed that sheets of electrodes placed on the cortical surface could turn thoughts into text or commands on a screen.
• In 2015 a quadriplegic used a brain interface to control an F-35 fighter in a computer simulation, while researchers funded by the Department of Defense have shown that users could remotely control a swarm of drones with their brains.
• BrainCo, a Massachusetts-based startup, makes non-invasive headbands that detect electrical brain signals and purport to indicate when students are in a state of concentration. "It's like a heart-rate monitor for your mind," says Max Newlon, BrainCo's president.

What's next: As BCI technology improves, it could enable truly transformative applications that merge humans and machines — especially for the military, as a report released this week by researchers at the RAND Corporation outlines.

• Cortically connecting a human brain to a machine could allow an AI to help a soldier in the field evaluate data far more rapidly than they could do alone, creating a "centaur" model combining humans and machines.
• DARPA has funded researchers who are studying the possibility of "synthetic telepathy," involving virtual communication on the battlefield through the analysis of neural signals.
• A tabletop exercise put on by RAND in 2018 was designed to see whether future BCI tech might be effective on the battlefield. "The answer was yes," says Timothy Marler, a senior research engineer at RAND.

The catch: For any of the more ambitious uses of BCI to become a reality, scientists will need to figure out how to implant connections in the brain that can last for a decade or longer — most current versions corrode in a few years — and that can function effectively outside a lab.

• BCI technology also carries the risk of malfunctions — and hacks by adversaries — as well as "the chance of relying too heavily on what you might term 'exquisite tech,'" says Anika Binnendijk, a political scientist at RAND.
• There are also major ethical concerns that will only grow as the technology improves — worst of all the "grave possibility that it would facilitate totalitarian control of humans," as the bioethicists Ellen McGee and G. Q. Maguire, Jr. noted more than two decades ago.
• That led a group of scientists working in the field to call in 2017 for a declaration of "neurorights" that would address the threats posed by brain-reading technology.

The bottom line: Like all the most important emerging technologies, BCI poses two questions: Can it be done? And should it be done? We shouldn't forget the second while figures like Musk focus on the first.