Guiding the autonomous flock: humans as AI shepherds
In this edition of the TechSonar, we continue to focus primarily on AI-related technologies, confidential computing being the exception.
This focus is driven by two key factors.
The first is the rapidly growing presence of AI-powered services in our everyday lives, alongside an expanding variety of innovative products and solutions that traverse diverse domains, encompassing online enterprises, educational institutions and the entertainment industry.
The second is the impact AI-enabled automation might have on fundamental rights.
This year's TechSonar report includes six trends: agentic AI, AI companions, automated proctoring, AI-driven personalised learning, coding assistants and confidential computing.
Agentic AI refers to artificial intelligence systems that can autonomously make decisions, take actions and achieve goals without constant human intervention. AI companions interact with and support humans through personalised experiences. Automated proctoring monitors online exams to detect cheating. AI-driven personalised learning customises content and learning experience to each student’s needs. Coding assistants help developers write and debug code. And confidential computing protects data while it is being used by performing computations in secure, isolated environments.
While each of these technologies serves a distinct purpose, they are deeply interconnected. Together, they illustrate how artificial intelligence is progressively reshaping not only business processes or common daily tasks, but also the human experience of technology.
Agentic AI provides the underlying autonomy that enables other AI systems, such as coding assistants and AI companions, to act with increasing initiative and contextual understanding. It also opens a new area in which AI systems built for different purposes and technologies can cooperate to achieve a common goal. Coding assistants, in turn, exemplify how these autonomous capabilities can augment human productivity, transforming the way software is conceived and maintained.
When artificial intelligence was first developed, the vision was clear: automate repetitive tasks to free humans for work that required empathy and connection. The rise of Agentic AI, AI-driven personalised learning and AI companions puts into question that vision. What was once seen as a tool for efficiency has become a platform for cognitive or emotional engagement, blurring the lines between automation and human connection. This blurring raises important questions about independence, trust and agency in human–machine interactions.
Automated proctoring stands at the intersection of these dynamics. It demonstrates the dual nature of AI deployment: while it can enhance integrity and efficiency in digital education, it also introduces new challenges in terms of transparency, fairness and proportionality of data use.
Finally, confidential computing connects all these developments by contributing to the technological foundation for trust. As AI systems increasingly rely on sensitive personal and contextual data, the ability to compute securely without exposing that data becomes central to ensuring privacy, accountability and compliance
Together, the 2025 TechSonar trends demonstrate how artificial intelligence is increasingly becoming part of daily life and expanding into areas involving more complex reasoning and deeper human-AI system interactions. These trends are shaping how we work, learn and relate to technology.
As AI systems grow more autonomous and deeply embedded in human environments, humans will increasingly be taking a role that we can describe as “shepherds of AI agents” - stepping back from doing tasks themselves and focusing instead on overseeing AI systems as they act, guiding their impact, and ensuring they align with human values.
The challenge will be twofold. First, we must ensure that technological progress continues to align with fundamental rights. Second, we must guarantee that the increase in AI autonomy does not result in a reduction of human agency, that is, people’s ability to make independent choices, exercise control over their actions and remain accountable for their decisions instead of having them dictated or constrained by algorithms. This will require targeted research and technical innovation, as well as an uncompromising commitment to supporting human values, thoughtful governance and collaboration across disciplines.
The EDPS will continue to monitor these developments and technology trends, fostering a dialogue that keeps privacy, accountability and human dignity at the heart of digital transformation.
Wojciech Wiewiórowski
European Data Protection Supervisor