AI writes all your code now, how will you ever debug it?
Greg Law & Mark Williamson
Kernighan said: everyone knows debugging is twice as hard as writing the code in the first place, so how will you ever debug it? With AI, it's now much worse than that. Today's AIs are much better at writing code than debugging it, and they still generate higher bug density than skilled humans.
We present our research into the state of the art of how good different LLMs are at debugging large, complex C++ code-bases, and we then show several tools and techniques to make it better.
Greg Law
Greg is co-founder and CEO at Undo.io. He is a coder at heart, but likes to bridge the gap between the business and software worlds.
Greg's background is in systems software development (operating systems, networking and tooling), and he has held development and management roles at companies including the pioneering British computer firm Acorn, as well as fast-growing start ups, NexWave and Solarflare. It was at Acorn that Greg met Julian and on evenings and weekends, they invented the core technology that would eventually become Undo.
Greg holds a PhD from City University, London and was nominated for the 2001 British Computer Society Distinguished Dissertation Award. He lives in Cambridge, UK with his wife Alison, their two children, two dogs and two cats. In his spare time, Greg catches up on email.
Mark Williamson
Mark started programming in the 80s, using Sinclair Basic on a ZX Spectrum + (the fancy one, with genuine plastic keys and 48KB of memory). Ever since then, he's wanted to explore what makes computer systems tick and build software that's useful to other people.
During his career, he's worked on high performance virtualisation, high performance storage protocols and smart material control systems (Google for Nitinol alloy, it's fascinating, spooky stuff!).
As CTO at Undo.io, Mark leads a team building the world's best time travel debugger - which is used to solve the hardest bugs in some of the world's most complex software.