OpenAI, from ChatGPT, says AI could add R$1 trillion to Brazil's GDP in three years

Artificial intelligence boosts productivity in companies The adoption of artificial intelligence (AI) could add up to R$986.7 billion to the Brazilian Gross Domestic Product (GDP) between 2027 and 2030, according to a Reglab study commissioned by OpenAI, the owner of ChatGPT. The value, calculated in present terms with a discount at the Selic rate of 13% per year, is equivalent to around 7.6% of the estimated GDP for 2026, of R$12.9 trillion. The survey estimates that AI can generate a cumulative impact equivalent to 2.77% of GDP over four years, with an average expansion of 0.69 percentage points per year in the economy resulting from technology. [bbc] Logo of the company OpenAI, responsible for ChatGPT Getty Images The study highlights, however, that gains should not appear all at once, but gradually, as companies, workers and governments adopt the new tools. The study was commissioned by OpenAI and estimates the impact of AI on 12 large sectors of the Brazilian economy and separates the effects of the technology into two channels: Increased worker productivity; and Greater efficiency of machines, equipment and production structures. Which area will benefit first? Transformation industries, those that convert raw materials and inputs into new products, concentrate the biggest impact, with an estimated R$264.2 billion between 2027 and 2030. Next come other service activities, with R$165.4 billion, and trade, with R$131.2 billion. According to the study, 65% of the total impact of AI would come from increased worker productivity, while the remaining 35% would be the result of greater capital efficiency. The proportion, however, changes significantly between sectors. "In agriculture, 92% of the total impact is captured via capital", points out the study. The sector could accumulate R$48.2 billion in impact, mainly through the incorporation of AI into machines, equipment and production infrastructure. Technological news, robots that impress the public at the largest artificial intelligence fair in the world, in Shanghai. Reproduction/TV Globo The effect of work, on the other hand, is predominant in sectors with a greater concentration of cognitive activities. In financial activities, for example, 87% of the estimated impact comes from increased worker productivity, the highest percentage among the 12 sectors analyzed. Banks, insurance companies and investment managers can use AI to review documents, assess risks, classify claims and serve customers. In the information and communications sector, which includes software developers, data centers and telecommunications operators, the estimated impact is R$49.3 billion, 97% of which is related to the labor factor. Economic impact of AI on sectoral production What matters is how the sector applies AI. The study highlights that the difference between sectors is related to the way in which AI can be incorporated into productive activities. In agriculture, for example, only 8% of the impact is attributed to the worker effect, because most rural activities are manual and have less cognitive exposure to technology. The research also differentiates exposure to AI and worker replacement. An occupation can have many tasks exposed to technology without this meaning that the entire function will be automated. In the central scenario adopted by the study, of every 100 tasks cognitively exposed to AI, 23 could pass the cost-benefit test for full automation, while the other 77 would enter the calculation as a complement to human productivity. "In practice, occupations rarely disappear completely: what AI reorganizes is the set of tasks within each role," the study states. Most of the work, therefore, would continue to be carried out by people, but with the support of technology. The survey cites as an example an administrative assistant, whose repetitive and codifiable activities - such as classifying documents, reconciling schedules and entering data - may be more susceptible to automation. A software engineer can have tasks accelerated by AI, such as code correction and documentation production, while judgment, coordination and project definition activities continue to depend on the professional. The "circular financing" system could have a domino effect if the AI ??bubble bursts Jaque Silva/NurPhoto/picture alliance Is Brazil on a good path? The incorporation of technology, however, must occur gradually. The study considers that the diffusion of AI in Brazil tends to be slower than in advanced economies due to heterogeneity between companies, the high cost of credit, gaps in professional qualifications and infrastructure limitations. According to projections, by 2030, 66.6% of the potential gains associated with increased worker productivity and 62.4% of the gains related to more productive machines, equipment and structures will have been realized. The study highlights that the R$986.7 billion represents an estimated potential, and not a guaranteed forecast. The materialization of gains will depend on the speed of technology adoption and the country's institutional, regulatory and educational conditions. The measures pointed out by the authors that can be adopted by companies are: Professional retraining for workers in activities with greater replacement potential; Expansion of digital infrastructure in the field; Incentives for the adoption of AI by companies with lower productivity; Implementation of a predictable regulatory framework; Governance and open data policies; and The use of technology in public services. "Thus, the materialization of these gains will depend not only on technological advancement itself, but also on the ability to create institutional, regulatory and educational conditions that favor its broad and productive diffusion", concludes the study.