OpenAI Codex Running Beta for Natural Language to Computer Code API
OpenAI has begun a beta test of its new tool for turning regular language into operational computer programming code. OpenAI Codex applies a version of the popular GPT-3 language model to instruct an AI in translating plain language instructions into a range of computer languages so there would be no need for programming training to build software.
Codex Lingo
As with GPT-3, Codex was built with an expansive dataset of public source code to teach the AI how to understand computer code as just another language. Though the AI can work with many languages like JavaScript, Perl, PHP, and more, OpenAI suggests that Python is where the tool is strongest right now. There are definitely some kinks still to work out, as OpenAI’s own researchers have found issues where it struggles to understand what people are saying and can even express religious and other prejudices if certain terms are not narrowly defined, as shown in a paper recently published by the developers. The beta is free for those invited to test it out, but will eventually be one of OpenAI’s paid services after the company feels it is ready for public use.
“We’ve created an improved version of OpenAI Codex, our AI system that translates natural language to code,” OpenAI explained in announcing the new program. “Proficient in more than a dozen programming languages, Codex can now interpret simple commands in natural language and execute them on the user’s behalf—making it possible to build a natural language interface to existing applications. We are now inviting businesses and developers to build on top of OpenAI Codex through our API.”
Copiloting Codex
As part of its deal with OpenAI for GPT-3, Microsoft has exclusive access to parts of Codex, but this beta is part of OpenAI’s plans for third-party developers. There’s also the new Github Copilot, still in previews, which is designed to get the AI to perform as a “pair programmer,” as when two developers simultaneously work on a coding project, commenting on each other’s work along the way. The AI is the junior partner and used Codex to understand what the programmer is doing and come up with suggestions. Copilot is supposed to be able to suggest solutions to problems, come up with tests, and even brainstorm features while tailoring itself to the user’s working methods. Copilot extends what Microsoft and OpenAI were already doing with GPT-3 and the low-code Power Apps programming tool, making it easier to program software with natural language.
Follow @voicebotai Follow @erichschwartz
OpenAI and GitHub Unveil New Copilot AI Assistant for Coding
Microsoft Scores Exclusive License to the Much-Hyped GPT-3 Language Model
