Jane Lee

Opinion | Why Hong Kong must embrace causal AI, the new reasoning intelligence

The analytical capability and transparency inherent in causal AI could give Hong Kong a solid advantage in AI deployment and governance

Reading Time:3 minutes

Why you can trust SCMP

Unlike traditional machine learning approaches that focus on identifying statistical patterns, causal AI focuses on understanding the causal relationships between variables in a system. Photo: Shutterstock

Jane Lee

Published: 9:30am, 21 Aug 2024Updated: 12:22pm, 21 Aug 2024

The discourse surrounding artificial intelligence (AI) has become increasingly complex as this rapidly advancing technology continues to transform industries, influence societal norms and raise profound ethical questions. As AI algorithms are applied across multiple domains, there are growing calls to ensure these systems are transparent, accountable and free of bias.

Questions have also been raised over the blurring of boundaries between humans and machines, as well as the ethical implications of machines with unbridled decision-making capabilities.

As Hong Kong explores AI governance, the inherent limitations and risks of current AI technologies underscore the growing importance of the emerging field of causal AI, also known as causal reasoning or modelling AI.

Unlike traditional machine learning approaches that focus on identifying statistical patterns, causal AI focuses on understanding the causal relationships between variables in a system, rather than associations or correlations, which are prone to data inaccuracies.

The field of causal AI is poised to play an increasingly important role in developing more trustworthy and valuable systems, and has shown impressive capabilities across domains. The methodology holds significant promise for high-stakes applications, such as in healthcare, finance and public policy, where “explainability” and accountability are paramount.

However, causal AI represents a paradigm shift in the human-AI dynamic and can help address the significant concern that increasingly, AI systems will have limited or no human interaction.