MindBullets 20 Years


First AI computer relies on sleeping students

Hermann Goethe is asleep at home. A dense network of electrodes is attached to his head and neck. It may look like some bizarre behavioral psychology experiment but Goethe is actually working.

“I get paid US$ 20 an hour while I sleep to connect my brain to Facebook’s Headcast,” says a smiling Goethe.

“We got tired of waiting for computers to get sophisticated enough to process data and become truly intelligent,” says Miriam Santiago, CMO of Headcast. “In October 2016 we managed to connect a network of fifteen sleeping engineers together and produced a computer capable of an equivalent 25 petaflops a second, more powerful than the US Energy Department’s Titan computer.”

At any one time some 25,000 sleeping students are connected up to Headcast. Facebook rents out blocks of processing time to governments, universities and corporations. “We limit our subjects’ brain utilization to about five percent, otherwise they wake up fatigued and disorientated. That still gives us about 4,100 petaflops per second to play with.”

And Headcast, what does it think of things? “It’s very enjoyable work. I solve computation problems on new disease vectors, product safety, economic value propositions, future forecasting for bank risk. I love it!”

ANALYSIS >> SYNTHESIS: How this scenario came to be

The field of Brain-Computer Interface (BCI) was first developed at the University of California Los Angeles in the 1970s. Implanted prostheses in the brain are, after adaptation, handled like a natural sensor or effector channel. The first human neuro-prosthetic devices implanted in humans were produced in the 1990s.

A number of companies have developed products utilizing BCI for a range of applications, from assisting the physically disabled, to enhancing computer game experiences.
2009: The promise of brain-computer interfaces
US/Australian firm Emotiv, demonstrates a non-invasive electroencephalography (EEG) headset which can read neural activity and interface with a computer-game.”Emotiv is a neuro-engineering company and we’ve created a brain computer interface that reads electrical impulses in the brain and translates them into commands that a video game can accept and control the game dynamically,” says Tan Le, president of Emotiv.

“This is the first headset that doesn’t require a large net of electrodes, or a technician to calibrate or operate it and doesn’t require gel on the scalp,” she says. “It also doesn’t cost tens of thousands of dollars.”

Dr Christopher James from the University of Southampton’s Institute of Sound and Vibration Research demonstrates the potential for people suffering from ‘locked-in’ syndrome to communicate, one letter at a time, via an experimental BCI system. IntendiX, a US-based firm, takes that experiment further with a commercial release at the CeBit 2010 show. The system costs US$ 12,000, pricey for many, but life-changing for those locked-in.

2012: Getting rid of the PC
“The slowest part of the interface is the computer. We have to tediously translate brain signals into computer signals, visualize these, then reverse the translation. What happens if we get rid of the computer and allow brains to communicate directly?” So asks Prof Richard Anderson from Caltech at the Annual BCI Research Award ceremony in 2012.

It is a strident call taken up by universities and research departments around the world. The US Defense Advanced Research Projects Agency (DARPA) promptly sets a US$ 1 million challenge for the first such communications system. “Can you imagine if we are able to communicate directly with troops without having them carry heavy computer displays around with them or radios, either of which can betray their presence?” asks Larry Kilroy, a DARPA spokesman.

Singapore’s Agency for Science, Technology and Research (A*STAR) focuses on assisting the recovery of stroke victims by directly stimulating appropriate nervous pathways through a process called operant conditioning. Such biofeedback offers potential to take up Anderson’s challenge.

However, the difficulties of establishing direct communication require the complete mapping of the human brain.

2013-2016: The Human Brain Project
The US government learns its lesson from the Human Genome Project. This time it will be a strictly competitive affair with the winner receiving US$ 1 billion in exchange for opening the information to the world. There is a catch. The research projects must be based in the US. The EU recognizes the threat as thousands of researchers leave European universities for the US, but the region is debt-ridden and cannot compete. Not so the Chinese, who authorize US$ 10 billion to establish a local innovation hub.

It looks like a replay of the Cold War: centralization vs the free market. Which will win?

It’s a close-run thing, but the US has a tremendous lead in world-class technical universities. MIT claims the prize in 2015, with China’s Shanghai BCI Institute only 10 months behind.

Google, Microsoft, Amazon and Facebook all offer hosted interfaces to the data and a surge of investment blossoms as venture capitalists flock to support new BCI projects. “We’re looking for the Twitter and Facebook of BCI,” says Colin Brookings of CapitalONE, a VC firm in Seattle.

Facebook is determined that this won’t happen. “We’ve got to ensure we are first to release direct friend updates to your mind,” says a spokesman. But the potential is more wide-ranging than that; what happens to mobile phones if you can interface directly?

2017: Headcast wants your brains
Facebook’s experimental project, Headcast, develops an interface that monitors the thoughts of fifteen of their engineers continuously. While they sleep, the Headcast software attempts to optimize its algorithms by getting its subjects’ brains to return the solutions to simple calculations.

“We were hoping that we would figure out how to allow a person to consciously ‘see’ the information we’re sending them. It’s incredibly difficult and we’re probably a decade away from figuring it out. However, we did figure out something else,” says Miriam Santiago, CMO of Headcast.

What the Facebook team created was a very powerful new form of distributed processing. “We couldn’t keep up with the number of calculations we could conduct while we were asleep,” says Hal Roberts, one of the engineers developing Headcast. The opportunity was obvious, develop tremendous new insight into how brains work by crowdsourcing the research and – simultaneously – create the world’s most powerful computer.

“The current incarnation of Headcast is by invitation only, but we’re expecting to enter a public beta in about a year,” says Santiago.

“I can’t wait,” says Headcast.

Warning: Hazardous thinking at work

Despite appearances to the contrary, Futureworld cannot and does not predict the future. Our Mindbullets scenarios are fictitious and designed purely to explore possible futures, challenge and stimulate strategic thinking. Use these at your own risk. Any reference to actual people, entities or events is entirely allegorical. Copyright Futureworld International Limited. Reproduction or distribution permitted only with recognition of Copyright and the inclusion of this disclaimer.