OpenAI develops a tool that translates English into programming codes

OpenAI, a company that studies the development of artificial intelligence, has created a new tool that allows you to translate English commands into lines of programming code. The function, called “Codex”, doesn’t even need literal terms to work and can be used even by those uninitiated in the field.

This is because, in the demonstrations presented by the company, Codex managed to create simple web pages and even more rudimentary minigames, just with phrases like “create a page with a side menu and a title in high”. The aim is to facilitate the work of programmers from different industries, as well as to encourage the study and entry of new professionals into this labor market.

Read also

“We see this as a tool to multiply programmers,” said Greg Brockman, co-founder and CTO of OpenAI. “There are two essential parts to programming: you have to ‘think through a problem well to understand it’ and also ‘map small pieces into existing code, whether it is a library, a function, or a problem. an API ”. This second part is the most tedious, because it needs people who are already programmers to do this “hard work” ”.

The Codex was built using GPT-3, the OpenAI language generation model, as a basis, and of which machine learning came from extensive training from various internet sources. GPT-3 was already able to generate code autonomously, but Codex goes much further, being specifically trained through open source repositories spread across the Internet.

OpenAI, however, intends to add monetary value to Codex: although the novelty is released as a free API, the company ultimately intends to charge for its access, in order to generate value for its investors. The company has said, however, that it does not want to use the tool to create its own programs.

Of course, being a tool still in its infancy, the Codex is not without its flaws: in a presentation on The Verge website, Brockman showed simple commands that were easily translated on screen – “add this image to ‘a person on the page and make it be controllable with the arrow keys of the keyboard’. Everything was working fine, until they realized that the “person” was continually leaving the field of view. The “Make sure person can’t leave page” command fixed this problem, but not without a cost: for some reason the image turned “flat” on the screen.

“There are times when she doesn’t know exactly what you asked for,” Brockman said. “So you have to think a little bit about what’s going on, but nothing that takes a lot of digging. “

There is no full release forecast for the tool yet, but Brockman points out that Microsoft has already shown interest in the technology: the CTO has mentioned to the company founded by Bill Gates the possibility of controlling pages. whole with words, generating interest in exploring the new tool.

I watched our new videos on Youtube? Subscribe to our channel!



Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Social Media

Most Popular

On Key

Related Posts