[Haskell-beginners] type inference: how ghc chooses e.g. between Float and Double?

Hi all, In ghci, let us type:

data Worker x y = Worker x y deriving Show let a = Worker 2 3.4 :t a a :: Worker Integer Double

According to which procedure does ghci chooses Integer for 2, and Double for 3.4? I believed that in such a case I was compelled to indicate a type signature explicitly, for example:

let a = Worker (2::Int) (3.4::Float) :t a a :: Worker Int Float

I knew that numbers in Haskell are polymorphic constants instances of the Num typeclass:

:t 2 2 :: Num a => a

But what is happening exactly when he chooses Integer and Double above? Thanks in advance, TP