Google is running a secret project to develop a neural network-based code generator that will be able to build, fix, and update code independently. This project is part of a larger corporate program in the area “Generative Artificial Intelligence”related to the creation of AI algorithms to generate images, videos and more.
According to reports, the code generator project codenamed Pitchfork originated from Alphabet’s X research division. This summer, the project was picked up by engineers from the Google Labs team, led by Olivia Hatalsky, who has been an employee of the X Division for many years and has been involved in the development of Google Glass and several other devices.
The source notes that the Pitchfork project is being implemented to do this “Teach code to write and rewrite itself”. It is assumed that the algorithm created will be able to learn programming styles and write code based on the lessons learned. Originally, the Pitchfork project was tasked with creating a tool that could update the Python programming language codebase to new versions without involving live programmers.
Over time, the project’s goals shifted toward creating a general-purpose system that could reduce the need for humans to write and update code while maintaining its quality. It is still unknown at what stage the development of the Google code generator is.