Supposing I give you a multivariate equation $$F\in\Bbb Z[x_1,\dots,x_n]$$
Following is undecidable:
'Is there an $(a_1,\dots,a_n)\in\Bbb N^n$ such that $F(a)=0$?'
However is the following always true?
'$\forall F$ that is non-trivial there are infinite $(a_1,\dots,a_n)\in\Bbb N^n$ such that $F(a)\neq 0$?'
It is true that there are infinitely many $n$-tuples $(a_1,\dots,a_n)$ of positive integers such that $P(a_1,\dots,a_n)\ne 0$. We outline a proof by induction on the number $n$ of variables.
For the case $n=1$, if the polynomial is a non-zero constant, there is nothing to prove. Otherwise, let the polynomial have degree $d\ge 1$. Then the term $a_dx^d$ is long-term dominant, so our polynomial is non-zero at any large enough natural number.
Now suppose that the result is true for polynomials in any number of variables less than $n$. We show it is true for polynomials in $n$ variables.
Let $P$ be a polynomial in the variables $x_1$ to $x_n$. Without loss of generality we may assume that $P$ mentions the variable $x_1$. Let $P$ be of degree $d\ge 1$ in $x_1$, and let coefficient of $x^d$ be $Q(x_2,\dots,x_n)$.
By the induction assumption, there exist natural numbers $a_2,\dots,a_n$ such that $Q(a_2,\dots, a_n)\ne 0$. Fix such an $(n-1)$-tuple.
Then $P(x_1,a_2,\dots,a_n)$ is a non-trivial polynomial in $1$ variable, so is nonzero for infinitely many positive integer values of $x_1$.