I used to be an advocate of 1 keyword per ad group, and while automation does make it manageable, I eventually abandoned that model and I will list a few reasons here:
Aggregating data – While you can aggregate data within your own database, you have to do extra work to configure the keyword data sets that you want to aggregate, so you are doing that work either in your ad groups or after the fact in your database reports. So you don’t save any labor, but you lose data because there are some things that Google will only report after you reach a threshold of volume. You get extra data from all of those aggregated groups that wouldn’t have reached the threshold when splitting into multiple ad groups.
Faster optimization – When you use some of the internal tools and features in AdWords, like optimizing better performing ads, the optimization occurs faster when nearly identical keywords are combined into the same ad group.
Simplification – Splitting and then re-aggregating data on the backend requires time to set up for each and every new report that you want to create. AdWords already provides the reports split by keywords so you are just doing work that has already be done. And again there is some data particularly within segments that would require quite a bit of work to set up and compile within the backend reports and it is already there if you have aggregated ad groups.
I do not believe that trying to include a specific number of keywords within an ad group is a good idea either. I only combine keywords into the same ad group when they are nearly identical in user intent and will have identical ads and landing pages. Many of my ad groups have only 2 or 3 keywords, while others might have as many as 30 or 40, it all depends on what works best for the campaign. I try to be as granular as I can be as long as it can improve results.
For me, it is not about what is easiest to manage, but what gives me the best end result, I am happy to do extra work if there is a payoff that merits the work.