Summary
Open AI is hiring developers around the globe that will write code to train its AI model, developers are responsible for writing code and explaining it to ChatGPT so that it can train on this new information.
OpenAI is a research organization that aims to create artificial intelligence (AI) systems that can benefit humanity, without being constrained by profit or power. One of its ambitious projects is to develop AI models that can understand and generate code, given natural language inputs. These models could potentially revolutionize software development, by making it easier, faster, and more accessible for anyone to create applications.
However, creating such models is not an easy task. It requires a lot of data, computation, and human expertise. That’s why OpenAI is hiring computer programmers to help improve its code generation models. In this post, we will explain what these programmers will do, and why their work is important for the future of AI and software.
Code generation is the process of automatically producing executable code from a high-level specification, such as natural language or graphical interface. Code generation can also work in the opposite direction: given a piece of code, a code generation model should be able to explain what it does in natural language. Code generation can also involve other tasks, such as transpiling code from one programming language to another, refactoring code to improve its quality or performance, or generating test cases or documentation for existing code.
OpenAI is investing heavily in making ChatGPT, its popular language model, better at coding. The company is hiring hundreds of contractors from different parts of the world to help create training data that will include lines of code and explanations of the code written in natural language. The ultimate goal is to improve Codex, a ChatGPT model designed for translating natural language into code.
Codex was launched in 2021 and was trained on data scraped from GitHub, the code repository owned by Microsoft. It’s good enough that Microsoft uses it to power GitHub Copilot, a service that helps programmers write code. However, recent research suggests that coding assistance could be improved by adding ChatGPT’s unique capacity for dialogue with humans.
Coding assistants such as Copilot are similar to cruise control in cars, where your hands need to remain on the wheel at all times because Copilot can generate errors in code with no awareness of the error. The question is, how do we get from systems that generate code that sometimes runs but sometimes doesn’t. And the answer to this is all of those systems today are not capable of planning; they are completely reactive.
OpenAI’s investment in hiring contractors to improve Codex is a step towards building a coding assistant that can anticipate the effects of its actions and build a mental model of the world. This would ultimately lead to more reliable and accurate code generation, and hopefully, a reduction in the amount of errors generated by current coding assistants.