After a brief welcome, Almassi set the stage by tracing the diverse histories that shaped the rise of generative AI, noting a pivotal moment in 2016 when algorithms gained the ability to learn with minimal human supervision.
Ethical issues
Afshin Almassi acknowledged that new challenges accompany generative AI and that there are valid concerns, particularly about ethics. The aim is clear: use data and advanced analytics, including generative AI, to improve public services while ensuring decisions positively affect citizens’ lives.
The expert panel explored the promises and hurdles of applying generative AI in the public sector, featuring five thought leaders: Javier Hernández Díez, deputy director general for the Promotion and Innovation of Digital Justice Services; Laura Mora Casado, Head of Analytical Services Area at the Social Security Data Office; José Miguel González Aguilera, deputy director general of Digital Services for the Autonomous Information Technologies Authority of Madrid City Council; Ignacio González García of the National Office of Fraud Investigation; and Ezequiel García Collantes, deputy director general for Artificial Intelligence and Digital Enabling Technologies at the Ministry of Economic Affairs and Digital Transformation.
The opening question asked how generative AI could enhance government services. Mora Casado highlighted a dual use within Social Security: speeding up internal processes while improving the user experience. She noted that this AI could enable individuals to perform complex procedures, such as applying for benefits, using data already on file about the user.
Improve accessibility
Hernández Díez elaborated that justice has long seen potential for AI and its implementation. Generative AI could translate legal language into clearer terms for the average citizen, with the priority of increasing accessibility and efficiency in justice.
García Collantes explained that the main challenge for the Ministry of Economy is to develop a data-rich solution for citizens. He described a chatbot that can personalize the experience using existing data, but emphasized that information exchange between ministries and administrations is highly complex.
González García cautioned that generative AI is now similar to how blockchain was viewed earlier: a solution seeking a problem. Governance faces very specific challenges, and while helping taxpayers matters, it is essential to avoid oversimplification because legal and procedural issues cannot be glossed over when deploying the technology.
Attempt
González Aguilera noted that Madrid City Council holds vast unstructured citizen information that could be mobilized to deliver services, with generative AI seen as another useful tool. He described the current moment in artificial intelligence as one of discovery.
The discussion also touched on issues such as bias in AI and the need for regulation at both national and European levels to govern its use responsibly.
García Collantes highlighted a major challenge: ensuring that AI aligns with Spanish law, given the legal system’s evolving nature and the difficulty of updating language models to prevent inaccurate outputs.
Great difficulties
Ethics and the goal of developing small, productive AI models capable of performing targeted tasks were raised, but experts agreed that society should first be informed so fear and hype are tempered. Mora Casado spoke vividly about the inevitability of generative AI and the need for conscious effort to shape its course: this technology cannot be stopped, but it must be guided thoughtfully.
Each expert wrapped up with key messages. González Aguilera urged applying AI at the local government level to improve citizen services. González García argued that the administration must adapt to society’s needs and lead responsible progress as citizens increasingly use artificial intelligence, aiming to prevent greater instability from unchecked use.
García Collantes offered practical guidance for incorporating generative AI into the public sector: stay ambitious and self-critical, avoid internal optimization alone, and remember the goal is to assist citizens. Talent and institutional support are essential to this effort.
Hernández Díez described the justice-focused objective as improving access to justice, while acknowledging significant challenges across legal, organizational, technical, investment, and especially communication and testing with concrete use cases.
Mora Casado underscored several themes, including data and social protection, and urged a bold but calm approach: eliminate public fear of AI and set an example. Almassi closed with a reminder that the central challenge of real AI deployment is keeping humans at the core, using the technology to deliver precise answers and improve human decision-making rather than merely making systems or infrastructures more efficient. Generative AI should be viewed as part of responsible analytics that transforms data into operational intelligence, enhancing services and decision-making in the public sector.