The point of problems like these is not that a long expression can be reduced to a short one. Instead, the idea is that expressions need to be well-defined, i.e., non-ambiguous. To attain such a goal, one introduces the notion of an "order of operations"; for you, this means BODMAS, while those in the U.S. are probably more familiar with PEMDAS.
One way to get this point across would be to present a single expression and talk about different ways you could interpret it (i.e., depending on the order in which you carry out the various operations). You could then enter the expression into a calculator several times. Note that the calculator gives you the same answer every time, and ask the student if he thinks the calculator will ever give a different answer.
It won't. Why? Because the calculator is programmed to follow a particular order of operations; that way, two people with the same computation to carry out will not somehow end up with different results (assuming that they don't make any errors of their own). Just as there is an order of operations implemented in the calculator, there is one to which we subscribe, as well.
But the reasoning is the same: to eliminate ambiguity.
You might go further and ask him about different values an expression can take on depending on where you put the brackets (or, as those in the U.S. would say, parentheses). Even for a single operation: Does it matter where the brackets go? This can lead naturally to a discussion of binary operations for which associativity holds. For example, $a+(b+c) = (a+b)+c$ for any $a,b,c \in \mathbb{Z}$; that is to say, for the set of integers, addition is an associative binary operation. But what happens when the operation is subtraction rather than addition? Similarly, $a\times(b\times c) = (a \times b)\times c$ for any $a,b,c \in \mathbb{Z}$; that is to say, for the set of integers, multiplication is an associative binary operation. But what about division?
As to the student's astute observation: Expressions can be written in many ways. For example, the relatively "short" expression $3^{3^{3}}$ has only three numbers, but is equal to $7,625,597,484,987$. (If you raise it to $3$ once more, then you get an obscenely large number.) Meanwhile, you could write $1 - 1$ a bunch of times with a $+$ in between each of them; the sum would then be $0$. So here we could have a "long" expression that, when evaluated, gives a short answer.
The salient point is that a well-defined expression for your purpose refers to exactly one number. There are many equivalent ways to write a single expression; some short, some long. In fact, our decimal notation is really a way of abbreviating, so that, e.g., $365 = (3\times 100) + (6 \times 10) + (5 \times 1)$. In fact, I could have written the previous expression without brackets, given that BODMAS makes it unambiguous.
As a final note: Students are often asked to "simplify" expressions. The idea of such exercises is to remove ambiguity, and, the hope is, it will be clear for any particular class and teacher what constitutes a correct simplification. For example, many teachers ask that students "rationalize denominators" and re-write something like $\displaystyle \frac{1}{\sqrt{2}}$ as $\displaystyle \frac{\sqrt{2}}{2}$, thereby ensuring the denominator is an integer. Generalizing far beyond your student's query, one might ask whether this notion of simplifying is itself unambiguous. You can find more on this question in a MathOverflow post here, though I mention this more for the curious reader than as a matter of responding to your student.
PEMDAS
is used...Parentheses, Exponents, Multiplication, Division, Addition, Subtraction. It's the same thing. $\endgroup$