博客

[FCE] British AI startup beats humans in international forecasting competition | Artificial intelligence (AI) | The Guardian

收听本期播客

阅读正文

In a striking breakthrough, ManticAI, a British artificial intelligence startup, has captured global attention by excelling in the Metaculus Cup, an international forecasting competition. Organized by a San Francisco-based company, the event challenged participants to predict the likelihood of 60 diverse events over the summer of 2025. These included political developments, such as changes in party leadership, and natural phenomena, like the scale of wildfires in the United States. Remarkably, ManticAI, co-founded by a former Google DeepMind researcher, secured eighth place, outperforming many human competitors, including seasoned professionals.

Forecasting, the art of predicting future events by analyzing data and trends, has long been a domain dominated by human experts who rely on experience and instinct. It plays a vital role in sectors like finance and business, where accurate predictions guide major decisions. ManticAI’s achievement highlights the growing potential of AI in this field. The system operates by dividing complex issues into smaller, manageable tasks, assigning them to specialized AI models. This approach enables it to evaluate current events, study historical data, and simulate various outcomes before delivering a prediction.

Although AI still falls short of the most skilled human forecasters, its rapid improvement has astonished many observers. Some industry experts predict that by 2029, AI could rival or even exceed the best human predictors. However, limitations persist. AI often struggles with intricate forecasts involving interconnected events, where maintaining logical consistency is essential. Nevertheless, its ability to process vast amounts of information tirelessly and handle multiple tasks simultaneously offers a significant edge over human forecasters in terms of speed and efficiency.

The outcome of the Metaculus Cup has sparked lively debate about the future of forecasting. Toby Shevlane, co-founder of ManticAI, believes that AI brings a unique perspective by avoiding the human tendency to conform to popular opinions. In contrast, some human forecasters argue that the ideal solution might lie in combining human intuition with AI’s analytical strength. As AI continues to evolve, finding a balance between technology and human judgment remains a critical challenge in this field. The success of ManticAI raises important questions about how industries will adapt to the increasing influence of artificial intelligence in decision-making processes.

阅读练习

1. What was the main achievement of ManticAI in the Metaculus Cup?

  • A. It won first place in the competition.
  • B. It outperformed many human forecasters by finishing eighth.
  • C. It predicted wildfires in the United States accurately.
  • D. It was the only AI to participate in the event.

2. How does ManticAI approach forecasting tasks?

  • A. By relying on human intuition and experience.
  • B. By focusing on a single AI model for all predictions.
  • C. By breaking down problems into smaller tasks for different AI models.
  • D. By ignoring historical data and focusing on current events.

3. What is one advantage of AI over human forecasters mentioned in the article?

  • A. It is always more accurate in complex predictions.
  • B. It can process information tirelessly and manage multiple tasks at once.
  • C. It relies on group opinions for better results.
  • D. It excels in maintaining logical consistency in all forecasts.

4. What does Toby Shevlane believe about AI in forecasting?

  • A. It will never match human forecasters.
  • B. It should completely replace human judgment.
  • C. It offers a fresh perspective by avoiding common human biases.
  • D. It is too limited to be useful in forecasting.

5. What does the article suggest about the future of forecasting?

  • A. AI will soon make human forecasters unnecessary.
  • B. Combining human intuition with AI’s strengths could be beneficial.
  • C. AI will struggle to improve beyond its current level.
  • D. Human forecasters will always be more reliable than AI.