Let $X$ be an affine variety over $k$ and $\overline{k}$ be the algebraic closure of $k$. Then the projection morphism $X_\overline{k} = X \times _k \overline{k} \to X$ has dimension 0 fibers.
Say $X = \text{Spec } A$. Then $X_\overline{k} = \text{Spec} (A \otimes _k \overline{k})$
The fiber of the homomorphism at point $p \in X$ is
$$X_p = X_\overline{k} \times _X \text{Spec} k(p) = \text{Spec} (A \otimes_k \overline{k} \otimes_A k(p)) = \text{Spec}(A_p/m_p A_p \otimes _k \overline{k}) $$
Thus, it is equivalent to say that the ring $A_p / m_p A_p \otimes_k \overline{k}$ has (Krull) dimension $0$. As far as I know, the tensor of two fields is not necessarily a field, thus ruling out any trivial dimension considerations. I am not really sure how to think of prime ideals in such a tensor product. Any ideas to tackle this would be appreciated.
An idea I'm formulating right now is that $k(p) = A_p / m_p A_p$ is a field extension of $k$. Thus the tensor is simply an extension of scalars, from one field to its algebraic closure. Is this perhaps equivalent to considering the same field $k(p)$ but now explicitly embedded in this larger field, namely the closure?
Since $k\to \bar k$ is integral so is $k(p)\to k(p)\otimes_k\bar k$ (Atiyah-Macdonald: Chap.5, Ex.3).
But then $\dim k(p)=\dim k(p)\otimes_k \bar k$ (Matsumura: Exer. 9.2), so that indeed $\dim k(p)\otimes_k \bar k=0$.
Edit: the fibers are actually finite
Since the ring $k(p)\otimes_k \bar k$ is noetherian of dimension zero, it is artinian and has thus finite discrete spectrum.
So all the fibers of $X_\overline{k} \to X$ are actually finite discrete schemes.
This was not obvious, because the scheme morphism itself $X_\overline{k} \to X$ is far from finite in general: think of $X=\operatorname {Spec}(k)=$ a point!