Brazil, due to its geographical position, scale, and level of adoption in application development, can lead the world in the use of artificial intelligence (AI). That is the message from Ronnie Chatterji, chief economist at OpenAI, one of the companies leading the exploration of this new technology worldwide.

“Brazil, with its unique combination of factors, could lead the world in this sector. It's a matter of putting the pieces together and ensuring that the public and private sectors work together judiciously,” he said in an exclusive interview with EXAME during his visit to Brazil in early October.

Chatterji, whose mission is to understand how AI will shape the global economy, unraveled the latest results from the first survey with external partners on how people use ChatGPT, the company's most famous product and synonymous with generative AI.

A professor of economics at Duke University and former advisor to the White House under Joe Biden, he also discussed how the use of AI will be reflected in the labor market and economic indicators—an effect that is still not very visible but is likely to take shape as more companies use the technology in their operations.

There is debate about whether generative AI could trigger a new productivity shock. Are we facing something similar to what happened in the 1990s with the internet revolution?

This is a fundamentally new technology that will have a huge impact on the way we live and work, and will appear in many of our economic statistics. As with the emergence of the internet, it takes time for these technologies to adapt to business. We are already seeing this in Brazil, where many companies are experimenting with AI. But they will need to figure out how to get a return on investment from AI tools. We see some doing this impressively, but it will become more widespread.

How will this be captured in economic data?

Part of it will not yet appear in the Gross Domestic Product, but it is the economic value we measure for consumers. When companies start using AI more and more, we will see it in GDP, and it will resemble previous waves of new technologies. I am optimistic that we are on this trajectory. But it will take time to appear in the data in Brazil and worldwide. AI will be like yeast as its use grows. It will help make workers more productive. I am thinking of those who spend a large part of their time on administrative tasks, which can be done using ChatGPT. That extra time can be used to devote to other things. It is less time for more production. That is how AI, in the beginning, will be that yeast you mentioned in terms of increasing productivity. But there is more.

What?

If we think about innovation that drives economic growth also in terms of total factor productivity. Innovation is the process of creating new ideas, whether in science or technology, and then scaling them up to commercial products. AI will be very important in discovering innovations. If we can accelerate the pace of innovation in Brazil and around the world, it will contribute significantly to economic growth.

In March, you launched a 12-month study to measure the impact of AI on the workforce. What have you found so far?

The number of users is growing very rapidly. Specifically for ChatGPT. It took two months for the number of users to reach 100 million. There are 700 million now. This is the fastest-growing consumer product we have ever seen, perhaps in history.

And we are finding that it is being used for three main use cases. We classified all messages in a recent article [“How People Use ChatGPT,” released in October] and tried to understand, for this sample of messages, what people were using it for.

In short, AI is being used to support decision-making. People have decisions to make, at work and at home, about how to spend their time, complete a project, or achieve their personal goals.

AI is being used as a co-pilot or co-worker to help people make these decisions. They often use it to ask a question, to help make a decision, to perform a task, or to express an idea or feeling.

These tend to be the most common use cases, more to complement humans and extend humanity than to replace people at work.

We need to observe this technology over time to understand its impact on jobs, especially in relation to replacement. Now, we see it as a complement, which is consistent with many economic models of technology adoption.

What other insights did this recent study bring?

We found that AI is being used to assist people with writing in a very important way. But people don't use it to write from scratch. They tend to use it in two-thirds of cases to edit or critique work. For me, this is important from a critical thinking perspective.

In addition, about 70% of the messages we are analyzing are for non-work-related use cases. We have not yet included corporate, programming, or API [application programming interface] use cases in this dataset.

Of the consumer data, 70% is not for work. It is quite surprising to see that, in a consumer product, 30% of usage is for work. It shows the flexibility of AI.

Repeating one of the most common questions you must hear: will AI steal our jobs? How do you view this question?

I tell people that it is much more likely that someone who knows how to use AI well will take your job than AI itself, because many of the human elements are becoming increasingly important in our work, even with the advancement of technology.

How so?

If you think about the work you and I do, much of it involves a human element, but there is a lot of groundwork involved in preparing discussions like this, writing articles, and doing research.

AI can help with many of these things and frees us up to do other, more human parts of our work that AI couldn't really do as well. But if we don't use AI, it's true that other people will use it to become more productive, to learn new skills, and to advance their goals.

That's why I tell people: instead of being afraid of AI and thinking about it in the context of losing your job, think about how it can improve yours. In addition, for the sectors most affected, we need to think about retraining and training.

Some recent articles suggest that AI has been taking over junior positions. Are you concerned that AI could create a gap in professional training?

Right now, there are many different impacts affecting the economy for what you call entry-level workers. One is rising interest rates around the world. Another is the impact of COVID-19, at least in the United States.

We have seen a situation where we are seeing less turnover in the economy—the number of people taking jobs and leaving jobs. And because many people who took jobs during the pandemic are leaving them at a lower rate than before, this is creating fewer opportunities for entry-level workers.

This is also happening amid the introduction of AI into the economy. So while these studies point to some potential impacts, we definitely need to do more research to find out if it's really AI that's making the difference.

Regardless, AI is being used in many companies, and some of these entry-level tasks are probably the easiest to apply to AI. This is where companies need to decide what their workforce will look like. If they don't hire enough young people, they won't be able to train their future executives.

How do you assess the use of AI in Brazil?

Traveling around the world and being here in Brazil, it's amazing to see the country as part of this group of nations that are really about to do great things. AI is already here: there are 50 million ChatGPT users. Most are young, seven out of ten users are under 35.

In addition to having many users, Brazil leads in developers. These are the programmers building applications based on it. So, I think of Brazil in the context of being a leader in the Southern Hemisphere. As a place that is using and learning to use these tools more effectively, but also developing applications.

The basic models that will be used by AI are a tremendous economic engine. But we will also drive growth by building applications based on AI. When I think about all the Brazilian developers who will have the ability to develop based on AI, that's what makes Brazil a really interesting place to look at the future of AI. I see it as a really interesting place for developers.

I see Brazil, just because of its geographic location, as having the ability to lead an entire region of the world. It all depends on the choices it makes in the future, but I think Brazil will be in a great position if it continues to leverage the value of AI.

In macroeconomic terms, how will AI boost the Brazilian economy?

Some sectors in which Brazil leads could be catalyzed by AI. In agribusiness, for example, one of the most important things is supply chain management. Getting goods from point A to point B at the right time to ensure that your inventories are not too high, but that there is enough stock to supply the buyer. If you can use AI in Brazil in your world-leading sector to make these supply chains work better, you will be more productive. These companies will grow faster, and this will increase Brazil's GDP over time.

What else?

In addition, the country has a large percentage of small businesses, which can grow with AI. One of the biggest challenges for a small business is: Can I access credit? Can I hire an employee? How can AI catalyze them for small businesses? Brazil is also in a very good position for this.

Finally, there is an ecosystem here to do many things that no one else can do. Not only the balance of the composition of economic sectors and their diversification, but also the fact that there are leading institutions with a stable and solid output of engineers and scientists, as well as sufficient land and energy resources to really drive the entire AI value chain.

Therefore, Brazil, with its unique combination, can lead the world in this sector. It is a matter of putting the pieces together and ensuring that the public and private sectors work together judiciously.