Agentic AI: The new technological challenge worrying experts
The artificial intelligence landscape has taken a significant turn. While tools like ChatGPT marked the beginning of this era, global attention is now focused on so-called agentic AI. Unlike traditional generative models, these new systems possess a superior capacity to execute complex tasks and make decisions autonomously, operating with increasingly reduced human supervision.
What defines agentic AI?
The main difference lies in their operability. While a conventional AI is limited to answering queries, an autonomous agent has the ability to chain actions to achieve a final goal. For example, in travel management, the system would not only search for information; it would be capable of comparing flights, synchronizing calendars, interacting with providers, and processing payments if it has the necessary authorizations.
- Execution capacity: Performs multiple steps to close a process without constant intervention.
- Strategic autonomy: Integrates into critical sectors such as banking, e-commerce, and insurance.
- Operational complexity: Its design allows for handling dynamic workflows in real time.
Risks in critical infrastructure and the dilemma of responsibility
Subject matter experts warn that the deployment of this technology in strategic areas, such as energy systems, healthcare, and defense, poses unprecedented risks. The primary concern is that, in its pursuit of optimizing results, an agent could irreversibly affect essential systems before a human can intervene.
Responsibility cannot be diluted under the excuse that artificial intelligence got out of control. We must demand that those who design and develop these tools assume the legal and ethical burden of their creations, just as it happens in other industrial sectors.
Nuria Oliver, director of the ELLIS Alicante Foundation
Towards international governance
The academic and technical debate is clear: it is not about halting progress, but about steering it toward transparent, safe, and controllable systems. Specialists such as Ramón López de Mántaras and Senén Barro agree that the narrative of a “rebellious AI” is a mistake, as it shifts responsibility from the developers to the technology itself.
To address this issue, Alofoke Deportes highlights the urgent need to establish international governance. This regulatory framework requires solid agreements between global technological powers to ensure that, before AI accesses ultra-sensitive sectors, there are evaluation mechanisms, clear limits, and a defined structure of accountability for any eventuality.
