There are a few different points to make here.
First of all, the notation "$\aleph_2$" is being horribly misused here (it reflects an implicit assumption of the generalized continuum hypothesis). Leaving that aside, the claim you want to prove is actually false. That is:
There are $2^{2^{\aleph_0}}$ (also denoted "$\beth_2$") many arbitrary functions from $\mathbb{R}$ to $\mathbb{R}$; by Cantor's theorem, $2^{2^{\aleph_0}}>2^{\aleph_0}$ and this latter (also called "$\beth_1$" or "$\mathfrak{c}$") is the cardinality of the continuum.
However, there are only $2^{\aleph_0}$ many continuous functions from $\mathbb{R}$ to $\mathbb{R}$. This is a standard exercise; one way to see it is to note that any continuous function $f:\mathbb{R}\rightarrow\mathbb{R}$ is determined entirely by the set $$S_f=\{(a,b,c)\in\mathbb{Q}^3: \vert f(a)-b\vert<c\}$$ (that is, if $f$ and $g$ are continuous then $f=g$ iff $S_f=S_g$) and then note that $\mathbb{Q}^3$ is countable so there only $2^{\aleph_0}$ many sets of rational triples at all (strictly speaking this only shows that there are at most $2^{\aleph_0}$ continuous functions from $\mathbb{R}$ to $\mathbb{R}$, but it's easy to show that's also a lower bound).
Finally, as to the step you mention, the point is that the meaning of cardinal exponentiation is exactly the size of the corresponding set of functions: $\kappa^\lambda$ is defined to be the cardinality of the set of functions from a set of cardinality $\lambda$ to a set of cardinality $\kappa$. So the set of all functions from reals to reals has cardinality $$(2^{\aleph_0})^{(2^{\aleph_0})}$$ since $\mathbb{R}$ has cardinality $2^{\aleph_0}$. Note that here we're looking at the set of all functions, not just the continuous ones.
It's then a basic result of set theory that the usual rule $(a^b)^c=a^{bc}$ holds for cardinal exponentiation, and that the product of two infinite cardinals is simply their maximum. So we get $$(2^{\aleph_0})^{2^{\aleph_0}}=2^{\aleph_0\cdot 2^{\aleph_0}}=2^{2^{\aleph_0}}$$ as Eric wrote.