Programming is one of the areas where large language models (LLMs) and new AI technologies are being implemented now. Here is a tool that one developer created to fix the wrong code elements.

Dev creates Wolverine AI tool to "regenerate" broken code elements

Last month, Twitter user BioBootloader showed a bug fixing tool called “Wolverine”. It was first reported by Hackaday and recently spotted by Ars Technica.

Just like the Marvel superhero, its gives Python scripts “regenerative healing abilities.” BioBootloader also posted a demo video showing how the Wolverine tool works.

“Run your scripts with it and when they crash, GPT-4 edits them and explains what went wrong,” the developer explained. “Even if you have many bugs it’ll repeatedly rerun until everything is fixed.”

So Wolverine runs the script, collects errors, and sends them to GPT-4 to figure out a solution. In the video, the tool came back with an explanation and suggested changes to the code. After that, the code is fixed and can run correctly.

Wolverine is currently available on GitHub. BioBootloader plans to further improve the tool by adding new features.

It is worth noting that Wolverine is just a proof of concept, but it is a great way to see how LLMs and AI technologies could help developers in the future. “Imagine running your tests through this to automatically fix any that you broke,” BioBootloader wrote.

Python is also not the most common programming language for game development, but some titles like Civilization use it to script game logic. Some developers have already started implementing AI into their games.

Some use Midjourney and Stable Diffusion to generate art. There is also an RPG called Tales of Syn, which utilizes these models and ChatGPT to create assets and dialogues. Last month, one developer also introduced a proof-of-concept tool to show how ChatGPT could be integrated into Unity.

Speaking of QA and bug fixing, Google recently patented the Gameplay Trainer system, an AI tech that should help developers automate the game testing process.


Got a story you'd like to share? Reach us at [email protected]

Tags: