News
This little neuron could be the key to artificial inteligence. Wrap your brain around that one.
Johanna Hung/the Gauntlet

Artificial intelligence at the U of C

Splicing neurons and micro-chips blurs the line between sci-fi and reality

Publication YearIssue Date 

While the fear of becoming cognitively inferior to an artificial intelligence manipulated brain-computer would have screamed 'science-fiction' a mere five years ago, today the groundwork exists to fuse human brain cells with microchips.

One such researcher works right here at the University of Calgary. Dr. Michael A. Colicos of the Hotchkiss Brain Institute is a neuroscientist, creator of the Virtual Intelligence Matrix, and avid eater of pistachios. His current work looks at the fusion of artificial intelligence and hybrid technologies.

You don't have to be a science fiction fan to appreciate the magnitude of excitement that was felt when the first team of scientists learned to grow neurons on a silicon chip.

This hybrid technology makes use of the two best information processing units in existence, making it potentially possible to exploit the cognitive abilities of the brain and the speed, talent for repetitive calculations and information storage capacity of computers, said Colicos.

One of the challenges faced when designing the interface was stimulating brain neurons without damaging them. This is the context in which Colicos, primary investigator in one of the two labs working on a neurosilicon interface here at the U of C, developed his photo- conductive stimulation technique.

By concurrently shining a laser on a neuron that's grown on a silicon wafer while applying a voltage difference across it, the current passes only through the illuminated area of least resistance, firing the cell specifically and non-invasively.

"Using this technology we have a really unique opportunity to quantitatively determine what structural changes in the wiring of neurons occur after specific information patterns are applied to them," said Colicos. "This helps us understand how the network works as a whole, and also how individual neurons process information."

Rather then fight crime by night like most people with double identities, Colicos also works on the Virtual Intelligence Matrix, an AI project which models how neurons process information. The VIM began where many great ideas are born: in a bar at three in the morning, when the idea for the Quantum Flux Engine­--a true random number generator based on quantum events--was born, said Colicos.

The QFE plays a critical role in the VIM, signifying the importance of randomness in Colicos's model of mental function, especially for simulating creativity.

"Randomness is definitely important, without a doubt," he said. "But the true nature of randomness is something that I can't say that I understand. The randomness in the computer program really only lasts for a split second at the start of a process."

"As the system instantly starts to form, the system itself starts to perturb the randomness and that is what makes it lose randomness. It's the loss of randomness that becomes the detection of information, that becomes meaningful."

The latest Quantum Flux program, VIWonder, is an internet based system that manipulates language-based intelligence.

"It takes a look at language and the order of words--how they appear in sentences," said Colicos. "It looks for and extracts patterns and associations, assimilating them into a relational matrix of ideas."

On a very superficial level, the program is like a webspider, he explained. The user enters a topic that they want to discuss and the information for the dialogue is then pulled from the net, carrying a train of thought. This information is presented as a conversation between two women pacing about in virtual reality.

Colicos plans to eventually fuse his work with the VIM with his work on the neuro-silicon interface.

"There are two major applications of the interfacing technology," he explained. "One is for medical research. For example, we're currently investigating cognitive dysfunctions like autism with our chips. But if we really perfect this interface, then you can have cultured neurons starting to do computational work. Mix that with artificial intelligence and you have the chance to really develop a sort of computer-assisted thought."

Despite the speed and power of computers these days, they still fall short of our brain when it comes to parallel processing. Living neuronal networks on the other hand, are known to be massively parallel, suitable for use as a central processing unit when tasks require major data processing capabilities.

"On one of our little wafers that we have, there are 100,000 neurons and those 100,000 neurons are all simultaneously working," said Colicos. "They can interconnect with each other and function in complete parallel. So, while it may eventually be possible to develop something like this the old-fashioned way, I think using neurons is the quickest way to achieve this goal."

Such an invention would have many exciting academic applications, said Colicos.

"There's no way one person can conceptualize all the information on [the national Library of Medicine database] these days, but there is a chance that a machine could do it," said Colicos. "If we can design a system which functions similar to how the brain works, then we can apply it to the research process while retaining creativity and all these kinds of features that the brain has. This would greatly accelerate our ability to generate good ideas and leads."

In response to Hollywood-tainted concerns that AI could potentially be used for evil purposes, Colicos stressed that advancements in AI could be accomplished ethically and without threat to the future of mankind.

"Skynet's coming, right?" joked Colicos, referring to the fictional AI computer network in the Terminator movies. "Although Hollywood finds it an excellent way to capitalize on fear, I think that it's going to be a technology that's very, very controlled. So no, I would not take a living neuronal-network-computer hybrid and put it in control of the missile defense system on earth--you don't even plug it into your printed if you don't have confidence in it."

"It is very important to have people who are terrified and throw up their arms and say 'Oh my god, what are you doing?' because that keeps the whole system in check," he said. "But you should never stop going ahead with technology, because it could be of great benefit."

Section: 

Issue: 

Comments

Itís definitely coming. The next intelligent being. Smarter, stronger, faster, more capable than us in (I have to say it) every way. The pace of technological advance is becoming staggering and itís only really starting to pick up. Actually I have to rephrase that last sentence. Itís still mind numbingly slow, compared to our progress once these ìnext generation computersî start to do a lot of the thinking for us or once they start to become more independent. Think of how much faster computer simulation has made development in some scientific fields and they are really just glorified spanners.
One thing I am wondering is how people will feel once they realize they are not the top dog anymore. Itís going to be a bit of a dint to our egos to be outdone by our own creation. Two options, restrict them or enhance us.

The way to prevent a massive AI uprising is to do the opposite of this research -- find a way to apply silicon to neurons. That is to say, find a way to augment the human brain with computer power instead of augmenting computer chips with neurons. I would absolutely pay tens of thousands of dollars for photographic memory, thought-controlled in-eye web browser or instant mind-to-mind communication. Once humans keep pace with computers, the next intelligent being will be an improved version of humanity and not some robot overlords.

Watch some Galaxy Express 999 for some insights about technological augmentation and humanity. Also, there are significant issues with Homo sapiens sapiens vs. Homo sapiens invictus vs. Homo sapiens astartes...