TechnologyOpenAI Creates a ChatGPT-Like Tool You Can Program: Why It Won't Replace Programmers, But Increase Demand for It

OpenAI Creates a ChatGPT-Like Tool You Can Program: Why It Won’t Replace Programmers, But Increase Demand for It

Open AI It’s on everyone’s lips these days. Your artificial intelligence text generation tool, ChatGPT, has sparked imaginations and led people to wonder what role AI will play in the future. The same company has also been working on Codexa less popular service, but one that could completely change the way programmers work.

Using the data it collects online, ChatGPT can answer almost any question—not always accurately—with an answer that looks like it was written by a human. Codex is similar, but instead of writing text, it writes source code. You can tell it what kind of software problem you’re trying to solve, and Codex suggests a solution in code form.

The fact that an artificial intelligence service can program almost as well as a programmer could raises 2 questions: on the one hand, How will this affect students? of informatics who aspire to dedicate themselves to this work tomorrow and, on the other, if these tools will make engineering profiles disappear well-paid software developers.

However, software development professionals can be expected to remain in high demand, as AI services like Codex will be a natural fit as programming becomes less complex, say several academics and computer education experts. .

ChatGPT is not only good for programming: it corrects code errors better than programs designed for it

ChatGPT, an artificial intelligence chat bot, has gone viral in the last two weeks.ChatGPT, an artificial intelligence chat bot, has gone viral in the last two weeks.

Codex has been available to programmers since 2021, as it is the program Copilot which was freely available through the Microsoft platform, GitHub. Copilot has been one of the reasons that have led OpenAI to receive several capital investments, including one of more than 10,000 million of dollars from the hands of Microsoft itself.

The company that runs sam altman has been investing even more in Codex recently: it has outsourced more than 1,000 professionals to help write code and set associated descriptions that allow Codex to become a better programmer, according to traffic light.

Codex and ChatGPT are a “huge productivity boost” for programmers, and many already report using them in their daily workflows, says Stanford University Professor of Computer Science and Linguistics Christopher Manning.

Still, Manning says that while Codex can program basic functions to make a developer’s life easier, that doesn’t mean it’s suddenly going to write entire applications on its own.

Each generation makes programming easier

When thinking about Codex, you have to understand that programming has become progressively easier with each generation of programmers, says Hadi Partovi, CEO and co-founder of the nonprofit educational platform Code.org, which develops curriculum. for computer classes taught in primary and secondary.

“The programing started with punch cards“, says Partovi. “Now we don’t use punch cards anymore.” Later, programmers began to write with keyboards using a programming language called Assemblya low-profile language that communicated directly with the architecture of the device in question.

Similarly, Codex further simplifies certain tasks performed by software engineers. The programmers they will not have to spend as much time memorizing as it’s been done millions of times, but they’re still going to have to understand the code that a tool like Codex generates, says Partovi.

Developers who use Codex or a similar tool and can’t explain what the source code it offers them does are clearly not going to be more productive just by using it. Codex can fill in lines and lines of source code, but programmers still have to understand—at least roughly—how to solve a technical problem.

“I’m sure that will make work easier for engineers“, says Partovi. “Which will make us have more engineers and more software engineering. The demand for technology is only limited by the supply of engineers.”

Building the next big revolution

For the next generation of programmers, one of the main concerns is that students will use tools like Copilot to write code and then feel bad that the program can do its job on its own, suggests the professor of computer science at the Stanford University, Cynthia Lee.

Lee says she’s already received student work that she’s pretty sure was done using Codex. Lee herself is concerned that the OpenAI technology could demotivate students who strive to solve their tasks by themselves.

Tools like Copilot are “a worsening of a problem that we have always faced, which is: ‘How do you get people to do the tasks they have to do so that they end up learning?'”, points out the professor.

One of China’s biggest tech giants plans to launch its own ChatGPT

A man takes a photo during a Baidu conference in BeijingA man takes a photo during a Baidu conference in Beijing

“You just have to have a lot of discussion with students about what’s really important: ‘Why are we here?'” Lee says.

Codex is a power multiplier that can speed up your programming work, but above all it is a bible that collects source code of software packages that developers have already written.

Even so, the Stanford professor is optimistic about this type of technology and stresses that what is important is that students continue to learn software development techniques. “There will always be a newly created barrier,” defends Lee.

Codex can accelerate innovation

One of the most important advantages that tools like Codex can bring is that may end up replacing manual searches what programmers sometimes have to do to figure out how to debug their code or find software packages that are compatible with the code they’re writing, explains Manning, the Stanford professor of computer science and linguistics.

For example, developers can use the programming language python to analyze the text of a web page. With Codex, they just have to write a comment asking for a code snippet to complete that task and the tool will return it to them.

“Even for those people who are doing this, the speed at which these models have improved and the success they have is quite amazing,” says Professor Manning.

“But these models are far from perfect and if you’re not able to tell when something is wrong and it’s generating bad code or there’s still something wrong with it, then you will not be more productive as a software engineer.

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Latest Posts

Read More
More