The Rise and Fall of Software Engineers

In the 1950s, being a software engineer looked something like this. Surrounded by huge machines and endless cables, these workers were truly rare to find at that time. According to a 1960 report from the United States Department of Labor, only 10,000 people across the country were engaged in programming. The scarce supply of these workers and the growing demand for technological services caused their salaries to increase exponentially over time. However, with the technological revolution of recent years, software engineers no longer seem so indispensable. It's becoming especially difficult for software developers to find jobs, and experts say that isn't changing anytime soon. In just the first months of 2025, nearly 18,000 tech workers have been replaced by automation technologies. AI can already do all of the jobs that we as humans do. This year, no mega cap is cutting as much and as quickly. And with the growing race for AI dominance, this trend does not seem likely to change. So how did the rise and fall of tech workers happen? In the 1950s, being a software engineer was nothing like it is today. Workers in this industry operated enormous computers that filled entire rooms. At that time, programming was not just writing lines of code. It involved punching cards to feed these rudimentary machines. No one imagined that, decades later, a machine would be capable of writing code on its own. More than a quarter of all new code at Google is now generated by AI. In those early years, languages like Fortran and COBOL marked the birth of programming. Programmers were a technical elite, and the process of creating software required a deep understanding of hardware architecture. As a result, there were so few programmers at the time that it is estimated fewer than 10,000 people across the United States had the knowledge and experience necessary to be considered software engineers. Back then, there was no concern about automation, as computers were slow and entirely dependent on human ingenuity. But as the 1970s and 1980s progressed, automation began to take shape. To these programmers, a computer presents a constantly changing image. Is it a good servant or a bad master? Compilers evolved, and the first integrated development environments emerged, making coding easier. In 1985, the release of Turbo Pascal revolutionized the market, allowing programmers to compile and execute programs efficiently. These advances reduced manual work, but creativity remained the exclusive domain of humans. By 1985, there were already about 200,000 programmers in the United States, and demand was rapidly increasing due to the rise of personal computers. PC sales are now exceeding business sales, up 25% last year and still accelerating. At the same time, the first attempts at automating specific tasks emerged, such as compiler optimization and static code analysis. However, the idea that a machine could replace an entire programmer still seemed distant. Software development remained a craft, with each line of code requiring meticulous attention. With the arrival of the new millennium, software development underwent a radical transformation. The first autocomplete and error detection programs appeared, streamlining repetitive tasks. Meanwhile, the open-source community flourished, with platforms like GitHub allowing developers to share and reuse code. During this period, programming remained labor-intensive, although automated tools began to ease the most tedious tasks. By 2010, the number of software developers worldwide exceeded 18 million. At the same time, more sophisticated code generation systems began to emerge. Tools like AutoML automated part of the machine learning process, making it easier to create models without advanced knowledge of statistics or mathematics. This development sparked both excitement and concern. If a machine could automate machine learning, what would stop it from doing the same with other areas of programming? But the real shift came in 2010 with the rise of artificial intelligence. AI poses a major threat to all of humanity. So could these programs really turn into science fiction like the Terminator movies, or are they a valuable tool that could benefit society? Advances in deep learning enabled the creation of models capable of analyzing vast amounts of data and detecting complex patterns. In 2015, Google introduced TensorFlow, a machine learning framework that marked a turning point in AI development. Tools like SonarQube emerged to automatically identify vulnerabilities and bad coding practices. In 2017, a paper published by Google introduced the Transformer architecture, the foundation of modern language models. This innovation allowed the training of AI models with natural language processing capabilities on an unprecedented scale. Over time, these models began to be applied to software development, generating code snippets and automating debugging tasks. But the most significant milestone came in 2021 when GitHub launched Copilot. Microsoft 365 Copilot. Copilot combines the power of large language models to turn your words into the most powerful productivity tool on the planet. Developed in collaboration with OpenAI, this system could write functional code from simple natural language descriptions. According to GitHub, in its first year of existence, Copilot generated approximately 40% of the code in the projects where it was used. This led developers to question whether their jobs were at risk. In 2022, DeepMind introduced AlphaCode, an AI system capable of solving competitive programming problems. Although it did not surpass the best human programmers, it ranked in the top 54% of coding competitions. This development made it clear that automation would not only affect simple tasks, but could also tackle complex problems once thought to be the exclusive domain of human thought. By 2025, AI is not just a support tool, it is an active collaborator. Language models like GPT-4 can generate complete applications from brief instructions, automating routine tasks. Companies like Microsoft and Amazon have already integrated AI into their workflows, reducing the need for junior programmers. Generative AI underpins everything we're doing here, with perceiving, grasping, and moving products with the Robin Arms, to managing fleet congestion with our Pegasus robotic drive. According to a report, up to 30% of software development tasks can now be automated by advanced AI. The consequences of this revolution are clear. Basic roles are disappearing as machines take over repetitive tasks. Companies prefer to use automated tools, rather than hire junior developers due to cost savings. On the other hand, the required skills are changing. Developers no longer just write code. They must understand how to interact with AI and validate its outputs. Looking to the future, projections indicate that AI will continue to advance. Emerging tools like DeepMind's AlphaCode have already demonstrated the ability to solve complex competitive programming problems However, opportunities are also evolving. Fields like cybersecurity, software architecture, and algorithm supervision remain areas where AI cannot yet replace human expertise. According to a study, by 2035, developers with hybrid skills combining AI and traditional programming will be in high demand. Moreover, creativity and innovation remain human strengths that AI cannot fully replicate. Developers exploring emerging fields like AI ethics, data privacy, and technology regulation have a vast arena for meaningful and sustainable careers. The displacement of software developers by AI is an ongoing process, but it is not an inevitable fate for everyone. Throughout history, automation has changed the nature of work, but also created new opportunities for those who adapt. Developers possess something that neither AI nor automation can match their capacity for reasoning and creativity. The future of software engineers will depend on their ability to adapt and work alongside artificial intelligence.