OpenAI makes a ChatGPT-like tool called Codex that can write software — Business d’Or

OpenAI has been all the rage lately because the company’s intelligent chatbot, ChatGPT, has sparked imaginations and asked people what role artificial intelligence will play in the future. The same company also developed Codex, a less popular service that could completely change the way developers work.

ChatGPT can answer almost any question you ask using data collected from all over the internet (but not always accurately). Codex is similar, except that you write software code instead of English sentences. You tell Codex what kind of software problem you’re trying to solve, and Codex uses code sequences to suggest solutions.

If AI services let developers write code, how will that affect computer science students and eliminate high-paying software engineering jobs? Yet software engineering skills remain high. In demand, according to academics and experts in computer science education, AI services like Codex are just a natural step up as programming becomes easier across generations.

Codex is now available to developers in the form of GitHub Copilot starting in 2021. And OpenAI, which has raised over $10 billion from Microsoft and other backers, recently increased its investment in Codex, hiring over 1,000 contractors to build code that will help Codex become better programmers. Semafor reports that it is preparing a description related to

Codex and ChatGPT are “great productivity gains” and many programmers are already using Codex in their daily workflow, says Christopher Manning, professor of computer science and linguistics at Stanford University.

Just because Codex can write basic functionality that makes an engineer’s life easier doesn’t mean it can suddenly write an entire application by itself, he added.

Every Generation Makes Coding Easier:

When you think of Codex, you need to understand that it gets easier to code with each generation, says CEO of nonprofit educational organization Code.org. Hadi Partovi, co-founder and co-founder, said. Curriculum for K-12 computer science classes, it says.

“Programming started with punch he cards,” he says Partovi. “I don’t use punch cards anymore.

Programmers then began typing at the keyboard, using a programming language called Assembly, a low-level language that communicated directly with the machine’s architecture. Programmers don’t have to spend that much time. Partovi says you can do the tedious work that others have done millions of times, but you can still do it with Codex Developers using Codex or similar tools can’t explain what the code is You need to understand the basics of how to solve problems. “We believe it will make engineering easier,” says Partovi.

Creating the next big thing:

For the next generation of programmers. My main concern is that students will be daunted coding with programs like Copilot. Cynthia Lee, a senior associate professor of computer science at Stanford University, says the program is based on the idea that it can do all the work on its own. She said she has already received assignments from students who she believes have completed them in the Codex.

Lee worries that the Codex will discourage students who are struggling to complete assignments. Tools like Copilot “exacerbate the problem we have always faced: how to get people to perform the tasks required to learn,” she said. rice field.

“We need to talk a lot with students about the real basics of why we are here,” Lee said.
Codex is a power multiplier that can speed up your programming efforts, but often spits out code that people have already written by gathering data from existing software packages. Still, Lee was overall optimistic about the technology and stressed the importance of students continuing to master software development skills. “Recreation always has its limits,” Lee said.

Codex Accelerates Innovation:

The advantage of tools like Codex is that developers can find ways to debug their code and find software packages that match those created on the web. It replaces the manual searches you normally have to do. Support the code that supports it. Write again, said Manning.

For example, a programmer can use her Python programming language to parse the text of her web page.

With Codex, just write a comment requesting code to perform that task, and the service will return it.

“Frankly, the speed and success with which these models have improved is astounding, even for people in the field,” said Manning. “But these models aren’t absolutely perfect. If you realize something is wrong and you generate the wrong code or still have errors, you can’t be a productive software developer. “

Originally published at https://businessdor.com on February 6, 2023.