Using concrete examples and insightful analysis, during the presentation “The Present and Future of Artificial Intelligence,” which took place on Wednesday, November 20, Mayr discussed the impact of this technology on key industries, the challenges it faces, and the trends that will shape its evolution in the coming years.
Mayr has Ph.D. in Computer Science from the Basic Sciences Education Program (Pedeciba), as well as has Master’s degree in Engineering (Research Track) and a Specialization Diploma in Big Data Analytics from Universidad ORT Uruguay. He is also a systems engineer and works as an independent consultant in AI and predictive modeling.
The Pillars of Modern AI
Mayr explained that the current development of artificial intelligence is based on three fundamental pillars: machine learning, reinforcement learning, and generative AI.
These areas are driving disruptive innovations in sectors such as healthcare, education, and software development, while transforming everyday tools.
“At least for me, Google is already taking a back seat. I’m using ChatGPT a lot,” Mayr said, illustrating how these technologies are changing the way we interact with information and solve problems.
Ethical and security challenges
In his presentation, Mayr emphasized that the advancement of AI is accompanied by significant ethical and security challenges, which he grouped into three main areas:
Threats to privacy
The widespread use of personal data to train models poses significant risks.
“Tools like Copilot have exposed private passwords from training databases, revealing sensitive information without users’ knowledge.”
Copilot is an AI-powered tool designed to assist programmers by automatically generating lines of code based on what the user types.
Data biases
AI models tend to replicate the biases present in the training data.
Mayr cited the example of predictive policing systems, tools that use AI to predict where crimes are most likely to occur, based on historical data.
However, these systems can reflect racial biases, as they tend to assign higher crime rates to neighborhoods with a high concentration of certain ethnic groups or demographic characteristics, thereby perpetuating existing inequalities rather than mitigating them.
The challenge of rigorous evaluation
Integrating AI into software requires a rethinking of testing and quality assurance methodologies.
Mayr pointed out that artificial intelligence models are not explicitly programmed like traditional software, but rather emerge from optimization processes based on large volumes of data.
This means that its behavior is not always predictable, which can lead to critical failures in scenarios not covered during training.
“The case of the Uber self-driving car that failed to detect a pedestrian coming around a corner shows that we need to adapt our evaluation methods to these new technologies,” he said.
The model used by the vehicle was designed to detect pedestrians only at intersections, an assumption that did not account for situations such as a pedestrian crossing in the middle of the street.
According to Mayr, this reveals an inherent weakness in AI systems: their ability to generalize depends on the data they were trained on and the conditions under which they were tested.
“When we integrate an AI-based system, we are incorporating a component whose operation may depend on variables that we do not fully control.”
For this reason, Mayr emphasized the need to rethink quality assurance and verification strategies in software development.
“Traditional testing methods aren’t enough. We need more robust and creative tests that simulate extreme or unexpected scenarios. Otherwise, we run the risk of serious failures in critical systems, such as self-driving cars or healthcare applications,” he warned.
Digital Transformation 2.0
For Mayr, AI should be viewed as a general-purpose technology that is redefining the boundaries of what can be automated.
He cited the "IDC 2024 AI Opportunities Study," released on November 12, which estimates a 3.7-fold return on investment for every dollar spent on AI and a 20% increase in the use of generative AI over the past year.
“AI is digital transformation 2.0. Everything that once seemed impossible to automate can now be absorbed by this new wave of technology,” he said.
https://www.youtube.com/watch?v=YnJb635f2Vk
Learn more about the new Master's in Artificial Intelligence