Genetic Programming has become a highly successful technique for generating complex solutions to difficult problems, so much so that the GP community now maintains a list of human-competitive performance by GP systems. However there is still wide scope for generating improvements on the current algorithms, particularly in the areas of premature convergence (ie the system loses genetic diversity before it has generated an acceptable solution) and bloat (the system may generate a much larger and more complex solution than is actually necessary for the problem). This seminar describes recent and current work in the School of Computer Science, University College, UNSW-ADFA, on mechanisms to promote diversity and control bloat.