Threats and opportunities of Artificial Intelligence*
When I started my Master’s degree in Artificial Intelligence at the Faculty of Science and Technology of NOVA University Lisbon in 1999/2000, I had my first contact with Machine Learning routines and neural networks. At that time, it was already clear that the future would be promising – the increase in computational processing capacity and memory would make the massification of automatic pattern recognition inevitable.
Among the techniques marked by this evolution are neural networks based on “backpropagation”, a gradient estimation method used to train neural networks. Although this algorithm was introduced in the 1970s, it only gained real prominence after an article published in 1986 by David Rumelhart, Geoffrey Hinton, and Ronald Williams. Interestingly, this innovation is still used today by the new Artificial Intelligences based on Deep Learning.
Backpropagation uses advanced mathematical calculations, based on differential calculus (regression), to adjust the values of the “perceptrons” – the network nodes that process information in both directions. In the final result, each input parameter is evaluated for correspondence with an output, allowing the neural network to recognize patterns with high precision. This innovation boosted unsupervised learning, which reached a new level with the emergence of generative AI and Large Language Models (LLMs), which now have billions of sensors or parameters, thus raising the capacity for learning and analyzing information to an unprecedented level.
However, despite the current dominance of deep neural networks (Deep Learning), it is essential to remember that AI is not limited to this approach. Historically, AI has encompassed various techniques, such as artificial life, computer vision, natural language, genetic, and evolutionary algorithms. Today, Machine Learning using neural networks has monopolized the most evident results of AI research.
AI in the job market
The rise of generative AI inevitably raises questions about its impact on the job market. Is this technology creating more opportunities or replacing professionals?
On the one hand, the threats
The fear that LLMs will replace workers is understandable. However, this transformation was already underway long before generative AI, driven by process automation. Supermarkets with automated checkouts, accountants replaced by tax software, online courses without teachers/trainers, robots in factories, warehouses, or pharmacies—all these changes show that the transition to a more automated labor market didn’t start with generative AI.
Of course, many people could lose their jobs, but as MIT professor Richard Bookstaber once said, “No man is better than a machine. And no machine is better than a man with a machine”. From this perspective, the only certainty we have is that those who learn to use these “machines” will always be ahead of the rest. The real threat, therefore, lies not in generative AI alone but in the combination of automation and advanced robotics, which can eliminate a significant number of repetitive and operational functions. However, the total replacement of human professionals is still far from a reality.
On the other hand, the opportunities
Although there is widespread concern that generative AI will replace a large percentage of professionals, I believe that automation will be primarily responsible for reducing some jobs and creating many others. Generative AI should be seen as a tool to increase productivity and allow workers to focus not on repetitive or routine tasks but on more creative and strategic activities.
In various areas, this technology has proved to be a valuable resource for allowing professionals to focus on higher value-added activities. And here, training (upskilling, reskilling, and newskilling) for generative AI is very important, as workers who refuse to use this type of tool will become increasingly unfit to perform jobs now and in the future.
AI is already benefiting several sectors
Regardless of the challenges, AI is proving its worth in several areas. In healthcare, it helps analyze tests, personalize treatments and perform minimally invasive surgeries. In software engineering, AI has been a valuable support in generating and optimizing code, automating data structures, and automatically checking for errors. In the legal sector, it speeds up the analysis of documents, the drafting of contracts, and the checking of case law. In research and science, it facilitates predictive modeling and the discovery of new perspectives that contribute to generating knowledge.
In human resources, it facilitates talent management and makes recruitment more accurate, improving the match between profiles and opportunities. In design and content creation, AI’s strengths are applied to generating optimized images, videos and texts for various purposes, including digital campaigns. In the agricultural sector, it has made it possible to improve crop monitoring and harvest forecasting, making production more sustainable. Information security strengthens digital protection by identifying threats and anticipating cyberattacks.
The contributions of generative AI are mainly manifested in interaction AI, where human supervision remains essential to guarantee productivity and quality. This AI should, therefore, be seen as an opportunity for evolution, as it expands the capacity of professionals and makes processes more agile and efficient, whether through more informed decisions, faster task execution, or automation of repetitive processes.
In the end, the real challenge lies not in adopting AI, but in redefining the role of human beings in the digital ecosystem. The competitive advantage will lie in collaborating with technology and adapting it to new market demands. In other words: the future will not be a contest between humans and machines, but an era of humans empowered by AI, which will pave the way for professions that don’t yet exist… perhaps because we can’t yet imagine them.
*This article was originally published in HiperSuper.
