Comment by causal

As agents get better at code we trust them to produce more of it. There are still bugs to find, but the haystack gets bigger.

So the number of bugs to find remains constant but the amount of code to review scales with the capability of the agent.