Software Verification in the Age of Artificial Intelligence

Why It Matters

AI‑driven verification can dramatically improve software reliability, but only if organizations treat AI as a collaborative tool and base decisions on current, evidence‑based evaluations, influencing talent strategy and product quality.

Key Takeaways

• AI should augment, not replace, software engineers across the lifecycle.
• Orchestrated AI agents can handle requirements, design, coding, testing, and proof.
• Distinguish L‑type (skill‑leveling) vs E‑type (skill‑enhancing) AI breakthroughs.
• Empirical, state‑of‑the‑art evaluations are essential before drawing conclusions.
• Historical “no‑code” promises failed to eliminate programming, only raised abstraction.

Summary

The webcast, presented by Bertrand Meyer, examined how artificial intelligence reshapes software verification and the broader engineering workflow, arguing that AI must serve as an assistant rather than a replacement for human programmers.

Meyer outlined an ecosystem of specialized AI agents—handling requirements gathering, architectural design, contract specification, code generation, test creation, proof generation, and DevOps deployment. He introduced the L‑type versus E‑type breakthrough framework to assess whether AI levels the playing field or amplifies existing expertise, and warned that premature assessments based on outdated or free tools can mislead.

Citing the viral “vibe‑coding” tweet and past promises from COBOL to low‑code platforms, Meyer emphasized that each wave raised abstraction without eradicating the need for skilled engineers. He quoted, "If you’re using a tool from last year, you’re not measuring today’s reality," underscoring the importance of up‑to‑date empirical studies.

The talk suggests enterprises should invest in orchestrated AI pipelines while retaining human oversight, and researchers must produce rigorous, state‑of‑the‑art benchmarks. The distinction between skill‑leveling and skill‑enhancing AI will shape hiring, training, and competitive advantage in the software industry.

Original Description

Title: Software Verification in the Age of Artificial Intelligence

Date: May 7, 2026

Duration: 1HR

ABSTRACT

The AI tsunami is transforming every aspect of software engineering. What does it hold in store for the world’s software and for the profession itself? Beyond the buzz, can "vibe coding" scale up to the production of the high-quality systems the world increasingly requires? How do these evolutions affect testing and, more generally, software verification?

The talk will address these questions and, more generally, analyze what software development will look like in the new, AI-rich world. It will particularly focus on the need for a modern form of software verification, supported by advanced tools and combining the best of dynamic techniques (tests) and static ones (proofs), all supported by AI agents. It will also draw the lessons for software developers and testers who are (legitimately) concerned about career strategies want to navigate the new development/AI landscape successfully.

SPEAKER

Bertrand Meyer

Professor at ETH Zurich and CTO of Eiffel Software and Recognyze AI

MODERATOR

Will Tracz

SIGSOFT Special Projects Coordinator