In an article in Nature Human Behaviour, the Professor of International and Cyber Security and co-authors urge a closer examination of how the GenAI discourse shapes research and policy.
As generative AI (GenAI) continues to revolutionise various sectors, it is projected to influence many aspects of society. In an article titled “We need to understand the effect of narratives about generative AI”, published this month in Nature Human Behaviour, Professor of International and Cyber Security Anita Gohdes and her co-authors find that public concerns surrounding the societal impact of GenAI are increasingly shaped by different narratives that can significantly influence research priorities and policy agendas.
“Understanding the origins and dynamics of these narratives is crucial to effectively addressing AI's actual impacts and ensuring a constructive discourse about its risks and potential,” explain Gohdes and her colleagues.
Escalating concerns: Four key narratives surrounding the AI discourse
The authors' analysis reveals a significant surge in AI-related news articles in 2023, largely driven by GenAI, amid rising concerns about its potential negative impacts. They find that 46% of GenAI articles were negative, while positive coverage declined, indicating a shift in media portrayal and the need to examine the GenAI discourse more closely. Four main narratives are identified in this context:
- The existential risk narrative warns that future AI advancements could create systems beyond human control, potentially leading to catastrophic outcomes, including human extinction.
- The effective accelerationist narrative supports rapid AI development, arguing that the benefits of solving global problems outweigh the risks, which they consider minimal.
- The real, immediate societal risks narrative focuses on current threats, such as deepfakes, AI's environmental impact, and unequal access to technology, prioritising these over speculative long-term risks.
- The balanced risks narrative advocates for integrated AI governance, addressing both existential and immediate risks to create comprehensive policies for harm reduction.
Authors call for more research as GenAI continues to evolve
Gohdes and her co-authors contend that these narratives may reflect various interests – commercial, political, social, or normative – and warn that dominant narratives can skew policy agendas and research priorities, often at the expense of marginalised voices and less-publicised concerns. They call for an empirically grounded analysis of these narratives, urging independent research to understand the social, political, and economic forces driving them. As GenAI evolves, the article suggests, so too will the narratives, making ongoing evaluation crucial for responsible AI discourse.
The article “We need to understand the effect of narratives about generative AI” was written by Fabrizio Gilardi, Atoosa Kasirzadeh, Abraham Bernstein, Steffen Staab, and Anita Gohdes. It was published on 21 October 2024 in Nature Human Behaviour.
You can access the full article here.
The Hertie School is not responsible for any content linked or referred to from these pages. Views expressed by the author/interviewee may not necessarily reflect the views and values of the Hertie School.