Ad hoc workloads is there to help with plan cache pollution. If you have multiple single use queries, a stub is saved instead of the entire plan. If a query comes along again and thinks the plan stub is appropriate, then the query will cause the full plan associated to that stub to be stored in the plan cache.

If you have a ton of one off types of queries - this can be helpful. The downside is that you may start seeing parameter sniffing and longer running queries.

If you decide to enable it - be prepared to disable it if performance starts degrading. And of course, the best option is to test it in a suitable environment first.