Please clarify the term "efficient" in the context of in-class member initializers

C.45: Don't define a default constructor that only initializes data members; use in-class member initializers instead (link) says:

Using in-class member initializers lets the compiler generate the function for you. The compiler-generated function can be more efficient.

Moreover C.48: Prefer in-class initializers to member initializers in constructors for constant initializers (link) says about in-class member initializers:

It leads to the shortest and most efficient code

What is meant by "efficient" in this context? Does it mean to say that the use of in-class initializers leads to making programs faster?

I'm asking specifically because an Abseil C++ Tip from https://abseil.io/tips/61 seems to say exactly the opposite! At the start of its conclusion:

Default member initializers won’t make your program any faster.

(Actually I asked the Abseil team to remove that sentence, but it's still there: https://github.com/abseil/abseil.github.io/pull/423)

I hope you can clarify this issue.

Start with the C.45 and C.48 sections in CppCoreGuidelines.md at the linked lines, then compare their wording with the cited Abseil Tip 61. Clarify what “efficient” means in these guidelines and resolve whether the wording should distinguish generated code, source code, or runtime performance.