Elon Musk says humans must become cyborgs to stay relevant. Is he right?

Humans must become cyborgs if they are to stay relevant in a future dominated by artificial intelligence. That was the warning from Tesla founder Elon Musk, speaking at an event in Dubai this weekend.

Musk argued that as artificial intelligence becomes more sophisticated, it will lead to mass unemployment. “There will be fewer and fewer jobs that a robot can’t do better,” he said at the World Government Summit.

If humans want to continue to add value to the economy, they must augment their capabilities through a “merger of biological intelligence and machine intelligence”. If we fail to do this, we’ll risk becoming “house cats” to artificial intelligence.

And so we enter the realm of brain-computer (or brain-machine) interfaces, which cut out sluggish communication middlemen such as typing and talking in favour of direct, lag-free interactions between our brains and external devices.

The theory is that with sufficient knowledge of the neural activity in the brain it will be possible to create “neuroprosthetics” that could allow us to communicate complex ideas telepathically or give us additional cognitive (extra memory) or sensory (night vision) abilities. Musk says he’s working on an injectable mesh-like “neural lace” that fits on your brain to give it digital computing capabilities.

So where does the science end and the science fiction start?

So far, brain-computer interfaces have been used for relatively simple tasks, mainly to restore motor control for paralyzed patients and enable communication for locked-in patients with brain injuries that prevent them from communicating verbally or gesturally.

These interfaces involve decoding brain signals from the surface of the skull through EEG or via implanted electrodes and then translating those signals into a motion command for a robot or cursor.

There has also been some progress made in the other direction: using external electrical signals to stimulate the brain. This happened last year with Nathan Copeland, a paraplegic man who was fitted with a prosthetic hand with two-way feedback, meaning he can not only control the hand but “feel” when it’s being touched.

Although medical applications are driving the research, there are also commercially available playthings that allow for novelties such as “mind controlled” drone racing.

Still, these are a long way from Elon Musk’s vision of symbiosis between man and machine, which would require a much more granular understanding of the brain network that goes beyond the basics of motor control to more complex cognitive faculties like language and metaphor.

“We have over 80bn neurons in the brain. Our tools currently give us access to an extremely small number of neurons. With prosthetics, we’re maybe talking about 100 neurons. We need higher bandwidth interfaces,” said Bryan Johnson, founder of Kernel, which aims to augment human intelligence with AI.

Professor Panagiotis Artemiadis of Arizona State University has been trying to get more bandwidth using a 128-electrode EEG cap to allow a human to control a swarm of flying robots with their brain. “We can already decode basic concepts like closing a hand or moving an elbow, but we can’t decode more complex behaviors,” he said.

He has created a system that allows for a single person to control the collaborative movement of multiple drones, for example making the flock move closer together so that it can fit through a narrow pass.

He is skeptical that the rise of AI will render humans irrelevant.

“We are building these machines to serve humans,” he said.

Miguel Nicolelis, who has built brain-controlled exoskeletons and a brain-to-brain interface that allowed a rat in the United States to use the senses of the other in Brazil, agrees.

Humans won’t become irrelevant until machines can replicate the human brain – something Nicolelis believes is not possible.

“The idea that digital machines no matter how hyper-connected, how powerful, will one day surpass human capacity is total baloney,” he told the Guardian.

Nicolelis argues that the brain – contrary to what Musk and Singularity proponents like Ray Kurzweil say – is not computable because human consciousness is the result of unpredictable, nonlinear interactions among billions of cells. “Our brains do not work in an algorithmic way and are not digital machines,” he said.

“It used to be annoying to see these kinds of statements, but now it’s becoming serious. It’s leading to mass hysteria.”

Nicolelis acknowledges that digital automation will lead to “serious unemployment” among people who perform certain “mundane functions” that can be replicated by machines. “But that doesn’t mean the human species will become obsolete.”

He agrees with Musk that if we can interface directly with machines we can produce a “quantum leap” in what digital infrastructure has produced today, but predicts that humans will retain ultimate control.

This contrasts with current automated systems, like autopilot, where the human is merely supervising the operation of a computer. Similarly, doctors are outsourcing the diagnosis of certain diseases to supercomputers. Under these circumstances human skills diminish and people become subservient to machines.

“I’m thinking about a future where we reverse this trend. We use brain-machine interfaces to enhance our ability to treat people, to improve our quality of life,” he said.

Better communication between humans and machines, particularly the transmission of emotional signals from humans, will be a powerful tool for building trust in automated systems, added Artemiadis.

For example, it would allow for humans to hand over control to an autonomous car with confidence. “It’s about making the machine more intuitive using brain signals to understand whether the human is distracted or tired.”

Columbia’s Paul Sajda agreed. “Rather than put us in a doomsday scenario, let’s look at how the relationship between humans and machines can evolve.”

He said that most people will be “scared to death” to sit behind the wheel of a driverless car, but if the AI were able to read our emotional state it could start to make predications about our desires and build trust.

Sajda described the mostly non-verbal communication between a team of six Navy Seals, which includes gestures, emotional cues and facial expressions as well as some dialogue.

“In the future it will be three humans and four robots,” he said. “How can they ensure there are these team dynamics that allow them to operate at the same level as a human squad? It has to do with trust between them.”

Mind-reading devices or implants are likely to introduce unprecedented privacy concerns. Sadja talks about the notion of freedom of thought as an extension of freedom of speech.

“All of a sudden what’s in your head can be expressed and communicated. One’s private thoughts are important to protect, I don’t think anybody – government or any company – should be charged with protecting them.”

It’s a concern shared by the University of Calgary’s Walter Glannon, who studies neuroethics.

“There is a risk of the microchips being hacked by third parties. This could interfere with the user’s intention to perform actions, violate privacy by extracting information from the chip,” he said.

As it stands, these risks are theoretical.

“We really first have to understand the network [of the brain] and how all of these processing units communicate with each other and interact with the world,” said Artemiadis. “We are really far away.”