- Google is working on a tool that teaches code to write and rewrite itself.
- The project was started in the company’s moonshot X unit and moved to Google Labs this year.
- It’s part of a larger push in the field of generative AI.
Google is working on a secret project that uses machine learning to train code to write, fix, and update.
This project is part of a broader push by Google toward “generative AI,” which uses algorithms to create images, videos, code, and more. This could have profound implications for the future of business and the developers who write code.
The project, which originated in Alphabet’s “X” research unit and was codenamed Pitchfork, moved to Google’s Labs group this summer, according to people familiar with the matter. By moving into Google, it signaled its increased importance to executives. Google Labs is pursuing long-term bets, including virtual and augmented reality projects.
Pitchfork is now part of a new group at Labs called the AI Developer Assistance team led by Olivia Hatalsky, a longtime X employee who has worked on Google Glass and several other moonshot projects. Hatalsky, who ran Pitchfork at X, moved to Labs when it migrated last summer.
Pitchfork was designed to “teach code to write and rewrite itself,” according to internal documents seen by Insider. The tool is designed to learn programming styles and write new code based on those learnings, according to people who know it and patents reviewed by Insider.
“The team works closely with the research team. They work together to explore different use cases to help developers,” a Google spokesperson said.
Pitchfork’s original goal was to create a tool that could update Google’s Python programming language codebase to a newer version, a Google spokesperson confirmed. “The idea was: how do you go from one version to another without hiring all these software engineers?” said a person familiar with the early stages of the project.
The project’s goals shifted over time to a general-purpose system that could further reduce the need for humans to write and update code, while maintaining code quality. In job postings for X late last year, Hatalsky said she worked on a team “building the future of software engineering.”
Employees who spoke with Insider did so on condition of anonymity because they were not authorized to speak to the press. Their identities are known to Insider.
The boom of generative AI
Google and other tech companies have already made great strides in “generative AI.”
GitHub, which is owned by Microsoft, launched a tool called Copilot that suggests code snippets and functions as developers type. Developers use Copilot to build up to 40% of their code, and GitHub expects that number to double within five years, Bloomberg reported earlier this month.
Google is working on several other AI code projects. Its subsidiary DeepMind has a system called AlphaCode that uses AI to generate code, but currently focuses on competitive coding or writing programs at a competitive level.
Google is also working on a tool similar to GitHub’s Copilot that uses machine learning to generate snippet suggestions as developers type. Google senior research director Douglas Eck said at an event in New York earlier this month that the tool had improved coding iteration times by 6% among Google employees who had it. used.
Google’s AI Developer Assistance program goes further by training systems to do more of the work themselves. The project is still in its early stages, and Google will still need to consider tricky ethical considerations about how these models are trained, such as bias and potentially copyright issues.
A class action lawsuit was filed against GitHub earlier this month, alleging that the Copilot tool committed “software privacy on an unprecedented scale” by using AI to replicate open-source code, reported The Rod.
Are you a Google employee and have more to share? Do you have any advice? Contact journalist Hugh Langley at firstname.lastname@example.org or on encrypted messaging apps Signal and Telegram at +1 (628) 228-1836.