A pretty short read, but food for thought (?). These authors argue that generating molecules by genetic algorithm (GA), randomly modifying or combining molecules and choosing the "best" ones iteratively, does at least as well as more complicated methods for molecule generation and optimization.
They ascribe this more to "baseline beating" behaviour in the literature than genetic algorithms being particularly wonderful, but the takeaway - that we should consider what advantage new methods in molecular generation have over GAs - seems fairly valid.