You describe what you need, wait a few seconds, and receive ready-made code. For those considering a career in technology, seeing artificial intelligence do this might raise a very real question: is it still worth spending years studying programming?
The concern deserves a careful response. AI can already perform development tasks, but this does not allow us to conclude that the entire profession will be replaced, much less set a date for this. The work involves understanding problems, defining requirements, integrating systems, testing solutions, and monitoring what happens after a software starts being used.
For those considering studying Computer Engineering, this discussion also helps to understand the scope of the degree. Programming is part of the course, as is the study of hardware, networks, and systems that allow code to be transformed into real-world applications.
The advancement of AI is changing the way we learn and work with these technologies. Understanding this change is more useful than choosing a career based on predictions about the disappearance of professions.
Where did the prediction that AI would replace developers by 2040 come from?
The reference to 2040 appears in a article published in 2017 by researchers affiliated with Oak Ridge National Laboratory in the United States.
The text discusses the possibility that, by that year, machines will be writing most of their own code, with the advancement of technologies such as artificial intelligence, machine learning, and natural language processing. The discussion considers, especially, the complexity of programming for different types of hardware.
This is different from stating that all developers will be replaced by 2040. The article presents a technological projection; it does not set an expiration date for the career nor guarantee that the market will remain the same until then.
There are several decisions to be made between developing a program and adopting it in a company. It's necessary to evaluate its reliability, maintenance costs, integration with existing systems, and the consequences of a failure.
Therefore, predictions about the capabilities of these tools need to be interpreted in light of what has actually been studied. Automated code generation can advance significantly and, at the same time, create new demands for professionals who develop and maintain systems.
What AI is already doing in software development.
Programming assistants can suggest code snippets and functions based on received instructions and available context. This support has applications in learning, writing solutions, and exploring alternatives.
Herself GitHub Copilot documentation, However, it advises that suggestions be reviewed and tested: the generated code may contain errors and vulnerabilities or may not correspond to the developer's intention.
Imagine a store that needs to update its discount system. A tool can help write the function responsible for the calculation. Before that, someone needs to clarify which products are included in the promotion, whether the benefits can be combined, and what happens to canceled or partially returned purchases.
If these rules are incomplete, it's possible to produce code that does exactly what was asked and still generates incorrect charges.
This example shows why development begins before programming. Part of the work involves figuring out what needs to be built, even when the initial request seems simple.
After delivery, other questions arise: Did the change affect any existing functions? Is the system still responding well? Can the team identify and correct errors? These questions help assess the quality of the result beyond the speed at which the code was written.
Using AI requires knowing how to evaluate what it delivers.
Adopting these tools does not imply unrestricted confidence in the answers. Stack Overflow developer survey By 2025, 51% of the responding professionals said they use AI tools daily.
At the same time, among the participants who responded regarding accuracy, distrust in the responses was greater than confidence: 46% versus 33%.
For those who are learning, this combination deserves attention. A well-presented answer can seem correct even when it contains a difficult-to-detect error.
Consider a programming exercise that involves organizing thousands of records. Two solutions might produce the same result with a small amount of data, but exhibit very different behaviors when the volume increases. Understanding data structures and algorithms helps students investigate this difference.
The same applies to reading an explanation. If the tool suggests a correction, the student needs to be able to follow the reasoning, test the hypothesis, and recognize when the answer does not correspond to the problem.
Learning to program remains necessary to evaluate, adapt, and maintain code produced with AI assistance. During training, a good practice is to try to solve the exercise, use the tool to discuss difficulties, and then explain the solution in your own words. If the student can only reproduce the answer, there is still content to learn.
Read also: Applications of quantum computing: how this technology can transform different sectors..
What changes for someone starting their career?
Those entering this field need to develop autonomy while learning to work with increasingly capable tools. This involves knowing how to use assistance without depending on it to understand each stage of a project.
A portfolio project can be an opportunity to practice this process. Instead of simply presenting the application working, it's worthwhile to document what problem it solves, how the data was organized, what tests were performed, and what needed to be corrected.
Think of a scheduling app. Besides creating the screens, the student can investigate what happens when two people try to book the same time, how to allow cancellations, and how to recover information after a failure. These are questions that bring the project closer to a real-world use case.
It's also worth noting where AI has helped and how its suggestions have been verified. This allows for clear discussions about technical decisions and demonstrates effective participation in development.
Undergraduate studies offer the opportunity to gradually build this experience through guidance, review, and projects of increasing complexity. Using this period to ask questions, experiment, and correct mistakes is part of professional preparation.
How is the technology market in Brazil?
The Brazilian IT market is projected to generate US$$67.8 billion in 2025, according to a report by ABES Based on its sector study, the entity reported growth of 18.51% of the total cost per capita (TP3T) that year. For 2026, the projection released was for a smaller expansion of 5.31% of the total cost per capita (TP3T).
The numbers indicate a significant sector, but they also show that the growth rate varies.
In the field of qualifications, a study released by Brasscom In 2025, it highlighted back-end development as one of the demands of that year and pointed to data, artificial intelligence, machine learning, and information security as important areas for the following years.
These indicators help to understand the sector, but they are not a guarantee of hiring. Revenue growth, the need for qualified professionals, and the availability of entry-level positions are different measures.
For those choosing a degree, it's worth considering what skills are available in the field of interest and how the course allows for their development. Internships, projects, and consistent practice help to align learning with the demands of the workplace.
Compensation also needs to be analyzed with this care. A survey of systems development analysts portrays a specific occupation, not the salary of all computer engineering graduates. Experience, location, responsibilities, and employment type make a difference in the comparison.
Why does Computer Engineering remain relevant in this discussion?
Software depends on a structure to function. It uses processing power and memory, exchanges information over the network, and can interact with sensors, equipment, and other systems.
A Computer Engineering It allows you to study these relationships. At ESEG College, for example, the program combines content from programming, computer architecture, networks and embedded computing, as well as artificial intelligence and data.
Imagine a project to identify defects in products using images. Beyond the AI model, it would be necessary to consider image capture, data quality, the time available for analysis, and the equipment that would perform the processing.
If the solution needed to work on a device with limited memory or an unstable connection, those limitations would be part of the design from the start.
This type of problem helps to understand the relationship between software and hardware. It also shows why knowing only the tool used to generate code may be insufficient to understand how the entire solution works.
For students, exploring different areas during their undergraduate studies is a way to discover their interests: developing applications, working with data, studying embedded systems, or investigating infrastructure and security, for example.
Read also: Computer Engineering at ESEG: what's it like?
What areas of knowledge deserve attention during undergraduate studies?
Keeping up with new tools is part of training, but it's important to set aside time for acquiring the knowledge that allows you to understand them and evaluate their use.
Programming, algorithms, and data structures They help to organize problems, compare solutions, and understand the behavior of a program. The practice needs to include reading, writing, and correcting code.
Mathematics and statistics They support the study of data and artificial intelligence. When working with a model, it is necessary to know how to interpret results and ask whether the evaluation performed represents the situation in which it will be used.
Computer architecture, networks, and operating systems They help investigate what happens when an application consumes too many resources, loses connection, or becomes slow. The problem perceived by the user is not always in the most obvious piece of code.
There are also skills developed through group work. Explaining a decision, listening to a question, and organizing tasks are concrete parts of a project. A student may master their part of the implementation and still need to learn how to integrate it with the work of their colleagues.
This knowledge is gained through practice. When developing an application, the student needs to make choices, deal with limitations, and verify whether the result meets the proposed objective. It is in this process that the content ceases to be isolated subjects in the curriculum.
How does the Computer Science course at ESEG address these areas?
The Computer Engineering program at ESEG College lasts ten semesters and is offered in São Paulo. The curriculum includes Fundamentals of Programming and Algorithms, Python for Data Science, Data Structures, Databases, and Computer Architecture and Organization.
Throughout the course, subjects such as Big Data, Cloud Computing, Artificial Intelligence 1 and 2, and Computer Security 1 and 2 are covered. The curriculum also includes Integrative Projects and Automation Projects. Check out the complete curriculum.
For those evaluating training, it's worth noting this sequence and how the fundamentals relate to the more advanced content. Studying artificial intelligence involves understanding the data, software, and infrastructure needed to make it operational.
Choosing a degree requires considering both an interest in technology and a willingness to study, test solutions, and learn continuously. Tools will change throughout one's career; professionals will need to keep up with these changes and possess the knowledge to make their own assessments.
Learn about the course of Computer Engineering at ESEG, Explore the curriculum and check out the admission options.
AI already performs programming tasks, but there is no proven date for the replacement of the entire profession. The impact on work depends on the evolution of the tools, their adoption by companies, and the responsibilities involved in each role.
Myth. This statement simplifies a projection Published in 2017, the article discusses the possibility of machines writing most of the code themselves by 2040, without stating that all developers will lose their jobs in that year.
