Do we want decisions that nobody can explain?</minimax:tool_call>: Do we want decisions that nobody can explain? (Do we want decisions that nobody can explain?)Photo by Rodolfo Baumann on Pexels
AI News

Do we want decisions that nobody can explain?</minimax:tool_call>: Do we want decisions that nobody can explain? (Do we want decisions that nobody can explain?)

SAPO Notícias2 September 2026 at 00:01

Portugal is testing automated systems with artificial intelligence agents in the Public Administration to process applications, detect fraud, organize service queues, and support decisions in areas such as Social Security, taxes, or immigration. The authorities promise greater speed, reduced paperwork, and less waiting time for citizens.

However, the article warns about a fundamental problem: many of these systems operate as "black boxes," meaning the decisions they make are not easily explainable or auditable. It is no longer about mere computers following orders, but autonomous agents that decide the next steps on their own.

The article questions whether society is prepared to delegate important decisions to systems that not even the technicians themselves can fully explain. The lack of transparency in these processes raises concerns about accountability and citizens' rights.

Related articles

AI News

Pedro Abrunhosa: "Artificial Intelligence steals. That's what it does"

Pedro Abrunhosa warned about the dangers of artificial intelligence, considering it "steals," in addition to highlighting concerns about discourse manipulation, the degradation of international law, and the social isolation caused by excessive use of social media. The artist defended concerts and festivals as essential spaces for community gathering.

Now02/09/26, 00:42
OpenAI lança modelo capaz de ciberataques mas com sistema que limita ações
AI News

OpenAI launches model capable of cyberattacks but with system that limits actions

OpenAI announced the launch of its new artificial intelligence model called Astra, capable of invading well-protected computer systems autonomously. To mitigate risks, the company will include a monitoring system that can interrupt the model's actions, seeking to balance advanced capabilities with security measures.

SAPO Notícias02/09/26, 00:32
AI News

Back to school with intelligence

This opinion article by Ana Rita Bessa addresses the integration of artificial intelligence in the school context, in the context of the return to classes. The author questions the delay in understanding how to use AI for the benefit of education, highlighting that, although inclusion is widely discussed, the full understanding of how to combine technology with what school is and should continue to be has not yet been achieved.

Público02/09/26, 00:30
AI News

Ethics in AI (6): Bias – Framework and a Realistic Democratic-Liberal Solution

Miguel S. Albergaria explores in this article the problem of algorithmic bias in artificial intelligence, presenting a conceptual framework for the topic and proposing technical procedures to mitigate these biases in a realistic manner compatible with democratic-liberal values.

SAPO Tek01/09/26, 19:14