The MAS recently offered insights into how AI and machine learning (ML) technologies are shaping the future of policy development.
According to The Register, Edward S Robinson, the deputy managing director for economic policy and chief economist at MAS, highlighted both the potential benefits and limitations of AI during his speech at the 2024 Advanced Workshop for Central Bankers, organised by the National University of Singapore.
Robinson acknowledged the instrumental role AI has played in enhancing policy decisions, citing examples such as identifying unusual financial transactions and generating dynamic inflation expectations from social media. He praised AI/ML’s ability to adapt models based on data, which could mimic human expert judgement and capture economic dynamics in innovative ways. These approaches, Robinson noted, have the potential to outperform traditional models in forecasting inflation and other economic indicators.
However, the deputy managing director also pointed to the inherent limitations of current AI technologies, especially large language models (LLMs). He raised concerns about the sensitivity of AI/ML models to parameter selection and the difficulty in interpreting the logic behind their outputs. Furthermore, he noted that LLMs often struggle with logical reasoning and mathematical tasks, challenging their ability to provide credible explanations for their predictions.
Despite these challenges, Robinson remains optimistic about the future role of AI in economic modelling. He suggested a balanced approach, integrating AI as a complementary tool alongside traditional macroeconomic models. This strategy could leverage AI’s strengths in data analysis while maintaining the theoretical rigor of established economic frameworks.
MAS’s current approach to economic modelling reflects this balance, incorporating new developments while preserving core theoretical foundations. Robinson’s vision for the future includes further integration of AI techniques to enhance the efficiency and accuracy of economic policy-making, provided there is careful oversight to ensure the technology’s reliability.
In summary, while AI holds promise for revolutionising economic modelling and policy development, MAS underscores the need for a cautious and balanced approach to its integration. The potential benefits of AI in enhancing economic forecasts and decision-making are significant, but the challenges, particularly in model sensitivity and interpretation, must be addressed to fully realise this technology’s potential.
Copyright © 2024 RegTech Analyst
Copyright © 2018 RegTech Analyst
