Kevin Korb presented Intelligent Systems, lecture 1. With a fair bit of general discussion we painted some fairly broad strokes in the initial lecture. It definitely initiated some thinking on the topic but I am still not 100% sure what the course work is going to be like. It was mentioned that the best reveal on course focus was the assignment titles:
- Problem solving, Logic (I really hope this does not include propositional logic1)
- Bayesian Networks (Looking forward to learning about these as it seems to be a major research area)
- Machine learning (I imagine this will tie in well with the neural networks subject)
There was some time spent discussing the definitive of intelligence. I would argue that it is simply an agent acting in a way that best achieves its defined goals. There was some mention of being able to change/update goals but this would be random for all agents unless there was a parent goal… In which case the previous definition still suffices. In regards to general intelligence, it does not seem very practical to spend many resources in developing. We already have 7 billion generally intelligent agents. Software and robotic agents which have specific goals and can execute them better than humans seem much more practical in my opinion.
We ran through some of the basic categories and problem domains in intelligent systems then had some discussion on Turing tests. Kevin mentioned that in the course work we would most likely do some prolog programming which sound like it will be very interesting. I am hoping that we get the chance to do some Python implementation of the search algorithms and Knowledge representation systems that we will be studying.