In my field AI governance, the most pressing challenge is to translate ethical principles into actionable technical implementation, and situate these practices on the common ground between different ethical narratives around the world. Over the past years, various organizations and governments around the world have released their own AI principles and guidances for regulations. And research have found that rather than being fundamentally different due to cultural preferences, the difference lies in the storytelling about why these principles matters. But these stories are entrenched and deeply embedded in different identities and aspirations. Foreseeably, it will be challenging to recognize differences while looking for common ground, and it will be challenging to transform stories into concrete realities.

One of the most critical ethical challenges we face today is the governance of autonomous AI agents—advanced systems capable of performing tasks traditionally carried out by humans. These AI agents can learn from and adapt to their environment, making decisions to achieve specific objectives without human intervention. As these technologies become increasingly sophisticated and pervasive, with the ability to learn, adapt, and make decisions in real-time, they have the potential to revolutionize industries across the globe. However, with this potential comes significant ethical concerns. Ensuring that autonomous AI is developed and deployed responsibly is essential to balance innovation with the principles of justice, human agency, accountability, transparency, and benevolence. Addressing these ethical considerations is not just a technical challenge but a societal imperative that will guide how we navigate the future of AI-driven systems.

The most pressing challenge in my view is that, today, we live in a world of interconnected global issues of great complexity, which interlocks existing problems and supercharges power dynamics. The gap between the increasingly interconnected complexity of our problems and our inability to make sense of them is the "wisdom gap". What is urgently needed, therefore, is finding an exploratory space to bridge this gap and, at The BRIDGE Lab, we strive to contribute to creating such a space through the approach of "value sensitive innovation".

The most pressing ethical challenge I see is communication between different social subsystems and specialized disciplines. As technology, and AI in particular, affects the whole of society, it is urgent that we learn to talk to each other, and even more so to listen and understand each other. We need to have a common and holistic understanding of terms like ethics, individual rights, fairness, trust, values, explicability or intuition, perception, imagination, creativity, hallucination, in order to translate them from natural language to computational language. The translation cannot only be done with a business-oriented understanding and mindset, where ethics often comes late in the process.

I believe the most pressing challenges in publication ethics are around peer review integrity and quality. From issues with equity, reviewer fatigue and competency, to AI-misuse, fake review and misconduct, the variables to contend with form a long list of potential threats to trust in science.

Generative AI is blurring the boundaries between human originality and synthetic creativity. This raise pressing ethical challenges related to notions of authorship, intellectual property, and the intrinsic value of human creativity and experience. Beyond this, it is also increasingly threatening the livelihood of artists and authors.

One major concern in AI recommendation systems is ensuring that they do not reinforce biases or contribute to excessive user addiction. When metrics like user views and clicks are set as optimization goals, algorithms may try to capture user attention, continuously pushing similar or exaggerated content or spreading anxiety. Even if the content itself is harmless, the resulting "information cocoons" pose significant risks to users, potentially leading to a narrow and homogenized worldview. Additionally, the logic behind these recommendation systems may encourage users to spend excessive amounts of fragmented time on less valuable content, which could have negative effects on their cognitive abilities.