TECH Signal 405
Software Should Work 2026 conference warns AI erodes software trust, advocates formal methods and better tools
↗
Illustration only Photo by Vista Wei on Unsplash
Ben Cornia's recap of the Software Should Work 2026 conference in Columbia, Missouri identifies three themes: software as a social contract that should be invisible to users, AI as a force degrading trust and increasing pressure on engineers, and the need to adopt better engineering tools like formal methods rather than accepting the status quo.
The conference articulates a growing practitioner concern that AI is simultaneously finding vulnerabilities, creating them, and becoming one itself, while eroding engineers' ability to reason about their own work. It also signals renewed interest in formal methods and proven engineering practices as counterweights to AI-driven development pressure and the assumption that large-company tooling is universally appropriate.
Written by elseif from the cluster below · every claim links back to a sourceThe three things worth knowing
Software operates as a social contract with users, who should never have to be cognizant of the software itself, yet current practices around metadata harvesting, paywalled features, and buggy systems violate that contract.
AI is breaking down trust in software as a product, in engineers' ability to reason about their work, and in inter-engineer relationships, with employers tracking token usage and teams competing internally to avoid layoffs.
Formal methods, type systems, memory-safe languages, and other established tools provide better scaffolding for problem-solving than over-reliance on AI-generated code, and just because something works for Google or Meta does not make it blessed for everyone.
THE CLUSTER