Talent Arena in Barcelona: Toward a Safe and Aligned Artificial Intelligence Model

Cidatum at the Talent Arena in Barcelona: Toward a Model of Safe and Aligned Artificial Intelligence

This week, Rodrigo Valdelvira, an AI and Automation Engineer on the Cidatum team, attended the Talent Arena in Barcelona, held at the Fira de Montjuïc, reaffirming our commitment to cutting-edge science and technological ethics.

Among the talks and workshops he was able to attend, the opening session by Sam Ramadori, Co-President and CEO of LawZero (an organization co-founded by AI pioneer Yoshua Bengio), stood out. In his presentation, “Our Roadmap to Safe AI,” Ramadori addressed a critical challenge: the need for language models (LLMs) to cease being unpredictable “black boxes” and to strictly align with human goals and the fundamental norms of our society.

As AI capabilities advance, undesirable behaviors such as deception or self-preservation emerge and must be mitigated by design. LawZero proposes “Scientist AI”: a transparent and reliable architecture that prioritizes truth and facts over the mere prediction of words.

At Cidatum, we share the view that greater security should not compromise technological potential. We will continue to collaborate and conduct research so that innovation in AI is synonymous with resilience and trust. This collaboration has been made possible thanks to the partnership agreement with the Government of La Rioja.