As per my view, there are multiple reasons for this:

1. The capability of GA to be implemented as a 'universal optimizer' that could be used for optimizing any type of problem belonging to different fields.

2. Simplicity and ease of implementation.

3.Proper balance between exploration and exploitation could be achieved by setting parameters properly.

4. Logical reasoning behind the use of operators like selection, crossover and mutation.

5. Mathematical or theoretical analysis in terms of schema theory or Markov chain models for the success of GA.

6. One of the pioneer evolutionary algorithms.

As per my view, there are multiple reasons for this:

1. The capability of GA to be implemented as a 'universal optimizer' that could be used for optimizing any type of problem belonging to different fields.

2. Simplicity and ease of implementation.

3.Proper balance between exploration and exploitation could be achieved by setting parameters properly.

4. Logical reasoning behind the use of operators like selection, crossover and mutation.

5. Mathematical or theoretical analysis in terms of schema theory or Markov chain models for the success of GA.

6. One of the pioneer evolutionary algorithms.

Previous answer is good I think -- worth adding that temporary fashion happens within scientific research just as elsewhere -- eg artificial intelligence/robotics now seems very fashionable after decades of scepticism and dismissal. But may not last...

Cheers

jim

Tremendously powerful technique that was modeled after biological evolution, which of course has tremendous efficacy and power to create complex adaptations. The GA will find elegant, multifactorial solutions to problems by hidden paths that are unknown to the researcher and are unlikely to be discovered by traditional engineering approaches.

See the work by John Holland, and also my "Rethinking Evolution" and Crossovers Generate Non-Random Recombinants Under Darwinian Selection"

I think Apoorva Mishra 's answer is a very good well rounded answer. I would like to focus on what I think is the major element in the popularity of the GA tool: simplicity.

GA's are usually easy to implement and easy to understand for any computer savvy researcher from any field. Just figure out your representation and your fitness function and you are good to go. They also tend to be easy to modify. You can tweak your system and add features without much hassle.

Of course another factor in the popularity of GAs is that they often work. But even not knowing what will work, "let's try a GA" sounds like a good idea because you can have it done quickly and you can play with it to try to make it work better if it doesn't work as well as you hoped.