Responsibility and Ethics: The Foundations of Trustworthy Artificial Intelligence

Experts at the IPN Summer School emphasized that Artificial Intelligence should be used with critical thinking, ethical responsibility, and high-quality data to ensure reliable results.

Artificial Intelligence (AI) should be approached with responsibility, ethical awareness, and critical thinking rather than blind trust, according to María del Pilar González Gil, Ph.D., a researcher at the National Institute of Astrophysics, Optics and Electronics (INAOE), during the Summer School on Computational Cognitive Science and Natural Language Processing, organized by the Research Center for Computing (CIC) of the Instituto Politécnico Nacional (IPN).

Presenting the lecture Artificial Intelligence Applications: Opportunities and Risks, González Gil reminded attendees that decision-making remains a human responsibility—not an algorithm's. While AI offers extraordinary capabilities, she emphasized that reliable results depend entirely on training systems with accurate, high-quality data.

Addressing participants in the auditorium of the Higher School of Computer Science (ESCOM), the researcher explained that one of the most common misconceptions is believing that AI always produces correct answers. Current models generate probabilistic outputs rather than absolute truths, and their responses can diverge significantly from reality when they are trained on incomplete or unreliable datasets. For that reason, she stressed the importance of developing AI systems based on trustworthy information.

González Gil also advocated for open access to knowledge. "As a researcher, my fundamental principle is that knowledge should be free and accessible. Everyone should have equal access to it," she said.

At the same time, she questioned the business practices of some major Artificial Intelligence companies that generate commercial profits by using third-party works and information without authorization. She noted that many AI platforms have been trained using millions of online texts, images, and other materials without obtaining permission from their creators, leading to numerous copyright lawsuits.

Emphasizing that technological innovation must respect both ethical principles and intellectual property rights, the INAOE researcher argued that it is inappropriate to profit from another person's knowledge without recognizing or compensating its creators.

She further explained that generative Artificial Intelligence produces text, images, and other content by relying on statistical probabilities, meaning that the quality and reliability of its outputs depend entirely on the data used during training.

The researcher also warned about bias in AI datasets, noting that many systems inadvertently reproduce existing social inequalities because they are trained on information that is incomplete or insufficiently representative.

In addition, she highlighted the environmental impact of large-scale AI systems. Because advanced models require massive data centers that consume significant amounts of electricity and water, sustainability should become an essential consideration in the development of future AI technologies.

High-Quality Data: The Cornerstone of Reliable AI

During her lecture, From Biomedical Signals to Artificial Intelligence, Blanca Tovar Corona, a researcher at the Interdisciplinary Professional Unit in Engineering and Advanced Technologies (UPIITA), emphasized that Artificial Intelligence is not inherently intelligent; rather, it learns exclusively from the data it receives.

For that reason, she described data quality as the single most important factor in building reliable AI systems, particularly in healthcare. She explained that medical datasets must be thoroughly documented, accurately labeled by specialists, free from noise, and representative of diverse patient populations, since any errors or biases can significantly compromise model performance.

Tovar Corona also pointed out that the greatest challenge is not designing sophisticated algorithms but creating dependable medical databases—a process that requires years of research, strict ethical protocols, close collaboration between engineers and healthcare professionals, and the participation of numerous hospitals and patients.

She concluded that only high-quality datasets can enable the development of Artificial Intelligence tools capable of providing meaningful support for clinical diagnosis and improving healthcare outcomes.