← Back to Guide
Tree of Thought (ToT)

Tree of Thought prompting expands reasoning into branching paths where the AI explores various possibilities and evaluates the most promising one, ideal for decision-making or optimization problems where multiple options need consideration to arrive at the best solution, and the analogy is like a tree growing branches in different directions, with the AI pruning the weak ones to focus on the strongest path forward.

Examples

User Prompt: Solve 8 queens puzzle. Explore branches for row 1: place queen in column 1, 2, or 3. Evaluate each. AI Response: Column 1 leads to conflict in row 2; column 2 better, continue branching.

User Prompt: Choose best route from A to B with traffic. Branch options: highway, city street, detour. Evaluate time. AI Response: Highway: 30 min but traffic; city: 40 min clear; detour: 35 min optimal.

User Prompt: Plan a meal with ingredients: chicken, rice, veggies. Branch recipes: stir-fry, soup, salad. Evaluate nutrition. AI Response: Stir-fry balanced; soup lighter; salad healthiest.

Advantages

  • Improves problem-solving for tasks with multiple options, finding optimal solutions efficiently.
  • Reduces prompt redundancy by evaluating branches in one go, enhancing decision-making accuracy.
  • Boosts efficiency in exploratory queries, like optimization or planning, by pruning suboptimal paths.