Introduction to AI - Week 11

Philosophical Issues

Dreyfus's Objections

Propose a model of human skill that:

Five Stages of Skill Acquisition


Five Stages of Skill Acquisition (Cont'd)


Five Stages of Skill Acquisition (Cont'd)


Five Stages of Skill Acquisition (Cont'd)


Five Stages of Skill Acquisition (Cont'd)


Chunks

Dreyfus' answer: chess grandmasters recognise and respond to whole positions, not component chunks. (most chess positions consist of several chunks)

Nota bene: 1997 a computer program could beat the world chess champion for the first time (Deep Thought vs Kasparov)


Five Stages of Skill Acquisition

 Skill Level  Components  Perspective  Decision  Commitment
-------- ------------- -------------
-------- ------------- -------------
 1. Novice  Context-free  None  Analytical  Detached
-------- ------------- -------------
 2.Advanced Beginner  Context-free & situational  None  Analytical  Detached
-------- ------------- -------------
 3. Competent  Context-free & situational  Chosen  Analytical  Detached understanding & deciding Involved in outcome
-------- ------------- -------------
 4. Proficient  Context-free & situational   Experienced  Analytical  Involved understanding Detached deciding
-------- ------------- -------------
 5. Expert  Context-free & situational  Experienced  Intuitive  Involved
-------- ------------- -------------

Expert Systems vs. Intuitive Expertise

Roger Shank (1984): The word "expert system" is loaded with a great deal more implied intelligence than is warranted by their actual level of sophistication

Dreyfus: "Actually, we'd prefer to call them "competent systems," since we can find no evidence that they will ever surpass the third stage of our skill model."

Feigenbaum: "The matters that set experts apart from beginners are symbolic, inferential and rooted in experiential knowledge. Human experts have acquired their expertise not only from explicit knowledge found in textbooks and lectures, but also from experience. ..." All the expert system builder need do is extract this knowledge and program them into a computer.


The Chinese Room Argument


The Lucas-Penrose Argument


But, Do the Arguments not apply to
Machines and Humans alike?

There are problems which are very hard, for machines and humans. We do not yet have a comprehensive understanding of intelligence, but the field is still very young and there is no convincing argument that it is impossible in principle to understand and implement intelligence.


Some Problems with Intelligent Software

Boeing 757, Birgenair 6.2.1996, "A Chain of Errors"
Still on the runway for the 80 knots test, the captain notices that the left speedometer does not work properly. He does not stop the start. Immediately after the start, the pilot switches the autopilot on. After two minutes, the board computer shows warnings: "Rudder ratio" & "Mach Speed Trim", shortly after that a warning: Machine too fast. However, at that point the machine was much too slow already. The autopilot changes the engines into neutral gear, the dynamic lift breaks away, the co-pilot changes into full speed ahead, the left engine breaks down, the aeroplane crashes in the ocean. In the end all 189 persons on board were dead.
           Die Zeit, 8.11.1996

1997: Boeing changes its general advice for the use of the autopilot in critical situations


What should a machine decide?

Should a pilot be allowed to enable thrust reverser when the aeroplane is still airborne?

Summary

Literature    



© Manfred Kerber, 2004, Introduction to AI
24.4.2005
The URL of this page is http://www.cs.bham.ac.uk/~mmk/Teaching/AI/Teaching/AI/l11.html.
URL of module http://www.cs.bham.ac.uk/~mmk/Teaching/AI/