Computing has accumulated significant technical debt over its 75-year history, exposing businesses and society to substantial risks. The use of AI-enabled engineering is amplifying both the benefits and costs. Researchers propose a pragmatic approach combining testing, specification, and proof to improve feedback loops for development. This involves co-developing executable specifications alongside code and tests, or using specifications that support various forms of testing and proof. However, practical implementation requires semantics infrastructure, including specifications and tooling for major programming languages. --- Why it matters: This matters because the proposed approach could significantly reduce technical debt and improve the reliability of software systems, which is crucial for AI development and deployment. Source: https://arxiv.org/abs/2608.19674