The final module in the first year of IGGI is over!
There's not too much to say about this one - it was another slightly weird module, this time looking at the idea of using AI players to playtest games, and then using the results to optimise the game design. It's still a relatively new area where there's lots of interesting research being performed, but I was a bit unsure as to who the module was aimed at.
Broadly speaking, the first IGGI cohort consists of two different sort of people: the people researching into conventional AI techniques, and people who are more worried about player experiences. The module didn't really click with members of the former group (myself included) as it ignores the fact that making "general AI players" for non-trivial games can be incredibly hard, so the idea that we can just throw AI techniques at games and get meaningful data back seems a bit over-simplified.
By contrast, the latter group of students seemed strongly against the idea that AI could be used in this way, instead arguing that games should be designed by humans, for humans. For many of their points I'd have to agree - AI agents can play games considerably differently to humans, and nice score distributions do not necessarily make a game fun - but there also seemed to be a bit of an anti-AI sentiment there. Understandable considering this was another module where MCTS was touted as the solution to all the world's ills!
Anyway, for two weeks I learnt about another area of research, programmed in Java for a bit, and even did an experiment to gauge players' enjoyment of variants of an Asteroids clone. Three things that were somewhat enjoyable at the time, but which I am going to run away from going forward.