Will an AI system be able to fully refactor a 10k+ line codebase before 2026 ?

Growing capabilities and context lenght increase of recent AI systems will potentially allow ever more powerful applications concerning code and IT infrastructures in general.

A full refactoring is a long and intense process that require a important amount of skill and knowledge. Good refactoring usually increase efficiency and readability of codebases, while facilitating further improvements on the codebase.

Refactoring & generation rules

• To be considered a valid refactoring, the AI refactoring should actually show, in one go : good readability, efficiency gain (if possible), harmonization of the syntax and structure of the code while not showing any loss in feature or specification.

• The system would need to deduce everything related to code, configuration files and basically the whole github repo

• Pre-generation user feedback is possible but should be 100% optionnal and should only concern architecture preferences, naming conventions and high level considerations.

• Re-run of the same input by the user until getting a valid result will not be counted as success.

Reliability

• It would need to have a very high average reliability (~95%+) accross various common programming languages (Python, Java, C++, C#, etc...) and librairies.

Allowed human interactions

• Interaction that need administrator privilege and directly asked by the system for package installation or similar for example (feedback possible for this).

Additionnal

• There is one attempt for the final code generation, but internally the system could go for as many iterative test-loop process as needed and use as many external tool as needed.

For resolution

I would prefer not to rely on a single source (including me) for the resolution,

that's why I will prefer using public benchmarks (that of course doesn't exist yet ...).

If not available I will go for online forum consensus.