It seems to me…
“As soon as it works, no one calls it AI anymore.” ~ John McCarthy.
Most of artificial intelligence’s (AI) accomplishments are under-appreciated. Current success in computer-based game playing, whether the 1997 Garry Kasparov vs. IBM’s Deep Blue chess or 2011 IBM Watson “Jeopardy” matches, can primarily be attributed to computing power rather than a form of artificial intelligence-based machine understanding. The average person has little comprehension of what AI actually is.
When most people think of AI, they usually assume it encompasses both awareness and consciousness; consequently tending to associate intelligence with creativity and original thought. While intelligence probably is achievable within the near future, independent originality will require significantly longer as relating the subjective experience to our objective universe is well beyond our current ability and is likely to remain so for quite some time. That said, AI applications are a rapidly evolving dynamic environment and have now progressed to where learning algorithms are complex and can self-adjust weights and rules in ways difficult to predict or understand afterwards why a device has taken some action.
The common perception of AI is primarily derived from science fiction where AI typically is cast as the villain. We currently exist in a symbiotic relationship with domestic animals and all types of devices and machines without any apparent conflict. Why should an even more sophisticated device than current computing systems be perceived as threatening?
Up until now, the only known method capable of producing fully autonomous as well as adaptive machines is biological evolution[i]. In non-embodied AI algorithms, intelligence is something that arises out of introspection. In contrast, evolutionary robotics supposes intelligence is able to arise out of ever more complex interactions between the machine and its environment. Metaheuristics (evolutionary algorithms) are used to optimize some or all aspects of an autonomous robot without assurance that the desired goal actually will evolve. While evolutionary roboticists believe in their approach to the development of autonomous robots, the field has yet to evolve a robot superior to one produced using mainstream optimization methods such as reinforcement learning. While there has been limited improvement in such areas as pattern and voice recognition, research currently being conducted involving thousands of parallel processors should demonstrate this approach’s potential within the next several years.
Ubiquitous computing where computation, communication, and sensing are enmeshed in the everyday world soon will be capable of seamlessly supporting us in everyday activities. Intelligence, rather than concentrated in any single location, will be distributed throughout the Web, for example, in the Internet of Things[ii] which concerns itself with four things: people, process, data, and things.
With the vast increase in information now becoming available, developments in machine intelligence and adaptive programming will permit computational processors to increasingly become cognizant of their physical surroundings and behave similarly to self-aware systems. One of the major challenges will be to understand and characterize the level of complexity in a world where countless numbers of devices are interacting with one another in unplanned ways[iii].
The Internet provides a cognitive extension to our mental limitations augmenting our basic abilities to quickly retrieve otherwise unavailable information. While computing will not make users more intelligent; it has a vital role in supporting thinking and problem analysis[iv].
That’s what I think, what about you?
[i] Bongard, Josh C. Evolutionary Robotics, Communications of the ACM, August 2013, pp74-83.
[ii] Bornmann, Lewis J., PhD. Internet of Things, https://lewbornmann.wordpress.com/2013/12/10/internet-of-things/, 10 December 2013.
[iii] Cerf, Vinton G. Cognitive Implants, Communications of the ACM, February 2014, p7, http://cacm.acm.org/magazines/2014/2/171677-cognitive-implants/fulltext.
[iv] Englebart, Douglas. Augmenting Human Intelligence: A Conceptual Framework, Stanford: Stanford Research Institute, http://www.dougengelbart.org/pubs/augment-3906.html, 1962.