User:CharlesGillingham/More/Strong AI

TODO
Strong AI


 * 1) Add Main:Artificial brain and move that material into artificial brain: make it two paragraphs (Yes)
 * 2) Summarize artificial consciousness stuff: make it two paragraphs. (Maybe)
 * 3) In Origin: Mention that "strong AI" is hard to construe and that some think it has bearing on the Dartmouth proposal. (Yes)
 * 4) Mention AI-completeness.

Why did early research into strong AI fail?
Many different reasons have been proposed for the failure of AI researchers to achieve strong AI within the time frames they expected. Some explanations are technical:

Lack of raw computing power and intractibility

 * Hans Moravec argued in 1976 that there was simply not enough computing power. Ray Kurzweil would later agree and both have made detailed calculations based on Moore's Law to determine when computer power would reach "human equivalence."
 * Sir James Lighthill noted that most of the algorithms in use suffered an intractable combinatorial explosion.

The problem is harder than anyone anticipated

 * Some, like Marvin Minsky and Douglas Lenat believed that researchers had seriously underestimated the complexity of commonsense reasoning and sheer amount of commonsense knowledge that strong AI would require.
 * Vision and robotics researchers like David Marr, Rodney Brooks and Hans Moravec believed that early successes in logic and chess were misleading because they were far easier than fundamental human skills like perception and motion, and speculated that higher functions like reasoning required these lower level skills.
 * Qualification problem
 * Unconscious reasoning
 * AI founder John McCarthy remained convinced that logic should be basis of AI, but had uncovered specific problems that had made progress difficult.

Looking in the wrong place

 * Many researchers criticized the symbolic, high level approaches of early AI programs and believed that computational intelligence paradigms like neural nets, fuzzy logic or genetic algorithms would eventually lead to strong AI.
 * On the other hand, Marvin Minsky would complain that by the 90s computational intelligence researchers had abandoned the central problems that needed to be solved for strong AI.

Other possible explanations came from philosophy and cognitive science:
 * Philosopher Hubert Dreyfus argued that most human skills did not require the kind of symbolic reasoning used by early AI programs at all, but instead used a kind of
 * Roger Penrose,  claimed that

Social reasons
And finally, there are the social explanations. In response to the lack of success, financing for strong AI research has been difficult to maintain.
 * In response to Lighthill's criticism and other factors, in the middle 70s DARPA and the British government cut off all funding that could be used for direct research into strong AI. Instead they chose to focus on projects with clear objectives, like autonomous tanks and battle management systems.
 * After the collapse of the billion dollar AI industry in the late 80s, many corporations and venture capitalists became reluctant to invest in AI projects that did not have some immediate commercial application in mind.