COMP6044   Just another ECS Blogs weblog

I’m currently reading Donald Gillies‘ “Artificial Intelligence and Scientific Method” as a introduction to AI.  The author has outlined inductivism and falsificationism as scientific methodologies relevant to the development and research of AI.

Issues concerning the topic are philosophical, psychological, logical and practical, and Gillies refers to the Turing machine as an example of the latter two, and to the program BACON.1 as psychological and logical.

Gillies has identified the development of expert systems as the first major break-through in the field of AI, although he notes the need for rule-based systems to solve the knowledge representation problem. The first expert system DENDRAL was essentially a chemist; the first expert system to (arguably) pass the Turing test (in a matter of speaking) was MYCIN, which had a knowledge base of some 400 rules.  The main stumbling block here is known as the “Feigenbaum bottleneck”.

I do have a question. Sir Francis Bacon wrote about scientific research which could/would/ought to be carried out “mechanically”. His examples include a mention of  a (circle-drawing) compass, which allows anyone to draw a perfect circle – something which by free-hand is near impossible, at least to most people. I find myself asking the question  – is using a compass to draw a circle a form of cognitive extension, even if it is, say, purely for fun, i.e. with no desire or intention to study the circle, use it to solve any problem, create a piece of art, etc. Does drawing a perfect circle with a compass = using a calculator to solve an equation?

It won’t be a surprise to anyone who sat in that class when we discussed our chosen topic that I found choosing really difficult. My problem fundamentally stemmed from one basic question: How can I examine A (the topic) through B and C (Disciplines 1 and 2) when I have no prior knowledge on any of them?  Add then to that the fact you’d hope the topics or disciplines had at least some relevance to something you might want to do in the future, or even just tickle your fancy…at least long enough for you not to immediately lose interest in all three. I was also afraid of my topic being too (for lack of a better word) ambitious, that the perspectives I was hoping to look at the topic from where too specialist or too wide, and I wouldn’t get a clear idea of anything. Yes, yes, that’s right, in other words, I was afraid of failing spectacularly.

After not-so-much brainstorming and much more just blurting out half-baked ideas, I have made the decision to go with a topic that I am very excited about. The two disciplines that I’m going to dip into are Psychology and AI. I’ve started doing some preliminary reading but nothing that I would confidently add to the bibliography as of yet…The most promising leads are from the seminar on Cognition and the Semantic Web I attended at Soton.ac.uk before term started, as well as some of the material we covered with Craig in his “Psychology 101” session last week in the coffee room. Having read some of the other blog posts as well, I’m hoping to chat to some of you in the next couple of days.