OpenAI wants ChatGPT to program itself

In the midst of massive layoffs from large companies, OpenAI, the company co-founded by Elon Musk and specialized in the development of artificial intelligence, has launched a wave of hiring with the aim of creating new AI models that will even improve its products. current ones, like ChatGPT.

According to traffic lightthe firm would have been recruiting more than 1,000 contractors distributed in different countries of the world so that, among other tasks, they can train artificial intelligence to be better at writing code.

600 of the 1,000 new workers that OpenAI has hired (40%) are programmers who will be in charge of producing a set of data to enable the AI ​​to perform software engineering tasks. The remaining 60% of new contractors, on the other hand, will be destined to carry out what in the field of artificial intelligence is called “data labeling”, which consists of creating raw data, such as images, text, audio or video , to later label them and thus specify their context to train the different artificial intelligence models.

Interestingly, a company job posting shared by an outside firm, details that “a well-established company” and “that is determined to provide world-class AI technology to make the world a better and more efficient place, is looking for a Python developer”, a high-level programming language.

OpenAI wants AI to be able to write code more accurately

One of the developers who applied for the position, and who did a 5-hour unpaid coding test, tells Semafor that OpenAI asked him to perform a series of tasks. Among them, explain in writing how you would solve a coding problem and carry out the solution. In the event that he found a flaw in the process, he also had to explain what the problem was and how it should be corrected.

The developer affirms to the aforementioned medium that “they most likely want to feed this model with a very specific type of training data, where the human being provides a step-by-step design of his thought process.”

The firm founded by Elon Musk, in fact, already has an artificial intelligence model capable of translating natural language into code called Codex. This, however, is limited in some cases. OpenAI, therefore, could be recruiting these workers to make the AI ​​capable of writing code more accurately. And, in this way, replace entry-level programmers in the future.

That the AI ​​capable of generating code is more intelligent, in addition, could serve to improve the current models that OpenAI has available, such as ChatGPT, in a faster way and thus be able to face other companies that also work on products based on artificial intelligence .