The term genetic algorithm can refer to the specific algorithm developed by John Holland in the 1970s, but is often used as a cover term for many different algorithms that all use an evolutionary process of repeatedly selecting a proportion of the best members of a population of solutions according to some specified criterion and using them to produce a new population of solutions with some chance of mutation and/or recombination. After repeating this procedure many times, the quality of solutions in the population tends to increase as judged by the selection criterion.
Evolutionary programming is what this technique is called when the evolving solutions can be interpreted as computer programs or functions, and this has consequences for the kinds of mutation and recombination operators can be used to modify solutions in the population.