The document discusses using artificial intelligence techniques like SBSE (Search-Based Software Engineering) to solve software engineering problems. It proposes that some objective functions used in SE are "surrogate" objectives that may be better learned than predefined, and that the goal should be passing a "Turing Test" of human judgment rather than optimizing a concrete metric. It issues a challenge to directly contribute AI/SBSE-generated code/patches to open source projects to get real-world feedback, rather than static interaction.