Natural Sources of Artificial Intelligence

This blog purports to begin a dialog between a neuroscientist and an interaction designer about the integration of interactive technology and the brain.  So what better topic to kick it off than the fusion of old fashioned artificial intelligence with the bloody real thing.

At least since the industrial revolution with its suggestive images of efficient robots performing mundane factory tasks without the melodrama of our dirty humanity, the popular conception of artificial intelligence and robotics has been one of cold, dispassionate creatures, made of either metal, or of thousands of lines of tightly structured computer code.  But the future of artificial intelligence is starting to look quite a bit wetter.

Computer artificial intelligence research and the design of “intelligent agents” have traditionally been heavily influenced by developments in neuroscientists’ understanding of brain processes, whether in terms of actual wiring of neural circuits, or on the more abstract symbolic level of cognitive processes.  Yet, despite what the computer scientists claim, we have still yet to see an artificially intelligent agent that does even the most mundane of tasks well, besides tasks of the most repetitive, inhuman sort.

But rather than crudely simulate the design of the human brain in intelligent agents, why not use the real thing?  Research done by Dr. Thomas DeMarse at the University of Florida recently grew brain tissue in a dish that learned to control a flight simulator program.

The intent of the research was to better understand the ways in which neurons communicate with one another.  In order to probe neuronal signaling pathways, the researchers grew the brain tissue on top of grid of electrodes that not only provided a sensory array with which to observe the electrical signals present within the brain tissue, but also allowed them to inject their own electrical signals into the brain tissue as a way of probing neuronal activity and providing feedback.

Dr. DeMarse did almost exactly what artificial intelligence programmers do, but using actual brain tissue instead of computer algorithms.  We are at a stage of understanding in both brain science and technology where such hybrid researches are becoming possible.

Will it one day be more efficient to use living tissue to perform complex computations?  Why use flawed, primitive computer models of the human mind when you can use the real thing?  If the cost of growing such natural sources of artificial intelligence were brought down, this could be a phenomenal area for growth (pun intended).