Many successful founders have taken the route to building the first version of the product without talking to a set of potential users.

On the other hand, there are many who have spent more time talking to potential users than they have on building the first version. They have been quite successful as well.

Recently, while reading the book, Mom's Test, I found a mental model of sorts to navigate through this.
The answer to the question that should we talk to a lot of users before start building or not depends on:

What are the risks one is solving for, Product Risks or Market Risks?

Market risks lie in our ability to understand the problem definition.

Product risks lie in our ability to come up with a solution to solve the identified problem.

If we are solving for Market risks, then we should talk to customers to understand the problem in terms of their context, Its priority, alternatives at hand, and evaluate business viability. 

If we are solving for Product risks, we should ship out an MVP(MLP now :) version that will help us get user's commitment on our version of the solution to the problem to get feedback via a system.

Solving product risks by talking to customers can make you run around in circles, similarly, solving market risks by building products can make you end up with too many first version products without any users.

Choose wisely!