Knowing how to program is no longer essential, says Amazon’s technology chief | Interview

Just as it is no longer necessary to know how to read maps to drive to new destinations or do math to share the payment for a dinner, the ability to program will no longer be essential to create apps or programs online – and that’s not necessarily a bad thing. This is the vision shared by Werner Vogels, the person responsible for technology at Amazon, during a conversation with PÚBLICO about changes in the technological sector and the impact of generative artificial intelligence (AI), which is the basis of ChatGPT, in the coming years.

For Vogels, the use of more AI assistants in programming is one of the main trends of 2024. The transformation of the educational model, to keep up with technology, is another. “We are going to have to rethink teaching”, argues the Dutchman. “With technological advances, people no longer need to know how to program to develop programs online. Creativity is enough”, explains the specialist, who has headed Amazon’s technology sector for two decades and is one of the main architects of AWS (Amazon Web Services), the arm of the company that offers services for creating and hosting websites It is apps.

Transforming programming into a more accessible task for everyone is a strategic front for technology. In 2023, AWS announced CodeWhisperer (something like “code whisperer”) that uses AI to suggest new blocks of computer code to programmers and fix errors as they create applications. The idea is to be able to do more and more, transforming ideas into code.

AWS is not alone in the race. One of CodeWhisperer’s main rivals is GitHub Copilot, a program developed by the Microsoft subsidiary that suggests lines of code based on GitHub’s extensive open source repository. There are also tools like OpenAI Codex and Tabnine.

Old concerns

With the trend comes concerns that thousands of programmers around the world will be left without jobs. A January analysis by the International Monetary Fund (IMF) indicates that 40% of jobs are already being affected – disappearing or changing significantly – due to artificial intelligence. The executive president of Nvidia, Jensen Huanghas also repeated the idea that programming will no longer be fundamental.

The person responsible for AWS sees the fears as part of an old narrative. “When you’ve been in the industry for a while, you see these concerns repeating themselves over and over again”, observes the Dutchman. The key is to “identify new opportunities”.

Vogels uses the case of the ice industry as an example: “Today, we have refrigerators at home. When they appeared, they killed the sale of huge blocks of ice for fresh food. Many people in the ice business lost their jobs, but new businesses emerged, because, with innovation, everyone could buy more food that didn’t spoil.”

An example of this is AWS’s commitment to using large language models, the basis of generative AI, to create tools that simplify the programming process. It is part of the DNA of the Amazon arm that was born when the company realized it could sell the technology it developed to become a commerce giant online.

“The fact that we have AI helping us with more and more tasks does not mean that we have ‘throw away’ critical thinking”, insists Dutchman Vogels.
Despite being fluent in English, the Amazon technologist admits that he frequently uses AI tools to review texts. “English continues to be my second language. I use many tools to correct and refine what I write,” he shares. “And I created a template with AWS tools to search through my old presentations and easily find a slide I used in the past.”

“We will always need engineers”, guarantees Vogels. “But more people will be able to work in IT and develop technology.”

Human at the center

Even if you stop programming directly, the human must remain involved in the process. “We have to remember that AI models don’t think. Models condense expressions that look good together in the context that the person gave”, notes Vogels. “The focus of primary education must be teaching people to think, reflect, adapt to situations and use their brain,” she explains. “You need to know how to reflect on mistakes and understand when a system doesn’t work well.”

Vogels recalls the scandal with the tax system in the Netherlands. In 2019, Dutch authorities were found to be using AI to detect and predict tax fraud. Thousands of people were forced to pay exorbitant amounts, by mistake, because the system considered that having dual nationality or low income was a risk factor. The problem took years to be detected and corrected.

It is one of several cases that motivated the European Union’s efforts to create regulation for the artificial intelligence sector, approved on March 13th (it will take some time for it to come into force).

“Machines can make predictions or make suggestions, but they cannot make decisions. This is the role of the human being”, highlights Vogel. “AI tools are precisely that, tools.”