Large language models (LLMs) are proving to be formidable contenders in the realm of future prediction, potentially surpassing human capabilities, according to a recent study by researchers from the London School of Economics, MIT, and the University of Pennsylvania.
A crucial aspect of economics and societal evolution may be effectively delegated to generative AI. Government policies, investment strategies, and global economic plans often hinge on accurate future predictions, a task traditionally performed by humans.
Peter S. Park, an AI existential safety postdoctoral fellow at MIT and coauthor of the study emphasizes the importance of accurate forecasting in various white-collar occupations such as law, business, and policy. The research indicates that just a dozen LLMs can match the predictive prowess of a team comprising 925 human forecasters.
Both LLMs and human forecasters were tasked with answering 31 yes-or-no questions predicting events three months into the future. Surprisingly, LLMs performed on par with human forecasters in terms of accuracy. In the second experiment, LLMs were provided with the median prediction from the human forecasters for each question, resulting in a significant improvement in prediction accuracy ranging from 17% to 28%.
Park expresses little surprise at these findings, citing historical trends that suggest ongoing advancements in AI cognitive capabilities. LLMs, trained on vast volumes of internet data, are adept at producing predictable and consensus-based responses, leveraging the collective wisdom of the crowd concept.
The implications of this study are profound, particularly for the future employment of human forecasters. As AI continues to demonstrate its predictive prowess, it may challenge traditional forecasting roles, potentially leading to shifts in employment dynamics.
The paper’s findings underscore the potential of AI to revolutionize the field of future prediction, raising questions about the role of humans in this domain. With LLMs capable of rivaling and even outperforming human forecasters, organizations may need to reassess their reliance on traditional forecasting methods. As businesses navigate this evolving landscape, initiatives such as the Best Workplaces for Innovators Awards recognize and celebrate companies fostering cultures of innovation.
