“Media development tools will gradually get features similar to ChatGPT”

It’s impossible to miss: since December, everyone has been talking about ChatGPT, a messaging software developed by the American company OpenAI. Founded in 2015 by a group of entrepreneurs and investors, including Elon Musk and Sam Altman, this California company aims to develop artificial intelligence (AI) technologies that “benefit all of humanity”.

A language machine is an artificial intelligence that is trained on large amounts of text and can process and understand natural human language. In fact, in July 2020 OpenAI launched the first version of its language engine, then called GPT-3. Although this novelty did not make much noise at the time and the use cases were still vague and theoretical, many people were impressed by its capabilities and understood that it represented a high level of innovation. .

ChatGPT, the failover

With ChatGPT, OpenAI goes further and makes the use of a language engine more concrete. Many people are enthusiastic about it. Others worry that robots could replace their jobs. Let’s face it: these technologies will not completely replace human work, but it is clear that they will bring changes in some professions, especially in the content and IT industry.

Journalism professions will experience changes because it is very easy to imagine the functions of editorial assistance, by helping to develop an article, by suggesting alternative formulations, titles or introductions for example, or even by easier verification of information. ChatGPT can actually generate text or code with an accuracy unmatched until now. All you have to do is give him a command and he will do it very well. On the internet, many examples published in recent weeks show that this technology has succeeded in writing a blog article, successfully answering an academic exam, summarizing a text, generating computer code, etc.

Towards a revolution in IT professions

Changes should be deeper for IT and web development applied to information. Media development tools – which are a type of CMS for writing code – will gradually acquire features similar to ChatGPT, and developers will need to master all aspects of it. Because if a language engine is not yet able to code on its own, in the long run, it can, for example, help a developer do it faster and better. With a simple instruction such as “Implement Subscribe With Google”, this simplified technical subscription module, a language engine will automatically write code that matches what was previously written.

To a lesser extent, this is already offered by GitHub Copilot, a tool co-developed by GitHub and OpenAI, which helps developers create computer code. Copilot is able to complete lines of code, create part of an application, correct errors, write test lines… At Le Monde, we had access to Copilot during its launch phase in March 2022 , and we were impressed with its capabilities. Development on Copilot gives the impression of programming with four hands, accompanied by a talented senior developer, when in fact it is a machine. Mastering these tools by technical teams will likely be more important than the ability to program manually.

Data science will continue to attract young people

A structural drawing movement. The craze for artificial intelligence is also spreading widely among computer buffs. It is possible that the latter will increasingly move to data science professions to contribute to the creation of these revolutionary AIs, instead of traditional programming professions, which may require fewer qualifications in the future. . This movement has already begun and is likely to accelerate. And although the developer profession still has a bright future, we can already see that data scientists are increasingly rare resources and sought after by the media.

Digital transformation of companies is becoming a necessity. We hope that this new technological revolution offered by artificial intelligence will allow less wealthy companies to invest in technology at a lower cost without having to acquire the necessary skills. This aspect is still uncertain.

ChatGPT is not without its imperfections, but it makes the capabilities of language engines more concrete and visible. It also allows everyone to imagine the future practices of these artificial intelligence technologies and others to come. GPT-4 was announced this year, and the biggest technology companies like Meta or Google are already working on alternatives.

External contribution