Google

This method involves breaking down a complex reasoning problem into several smaller sub-problems. Each sub-problem is then addressed individually, often in separate interactions with an LLM or by using other specialized systems, before the final solution is synthesized.

Multi-Run Problem Decomposition for Complex Reasoning

A user asks an AI assistant: 'What was the highest-grossing movie of 2019, and what was its main theme?' To answer this complex query by breaking it down into smaller, sequential tasks, arrange the following tasks in the most logical order.

An AI system is tasked with planning a multi-day hiking trip. This requires several distinct steps: checking weather forecasts from a live data source, finding available campsites through an external booking system, calculating the total distance and elevation gain using a mapping tool, and finally, creating a day-by-day itinerary that synthesizes all this information. Given the complexity and reliance on external, specialized tools, which of the following approaches is most likely to produce a 

Learn Before

Related