Chain-of-Thought Prompting
A prompting technique that improves AI reasoning by instructing the model to break complex problems into intermediate steps before arriving at a final answer.
Chain-of-thought (CoT) prompting, introduced by Google researchers in 2022, dramatically improved LLM performance on math, logic, and multi-step reasoning tasks. On the GSM8K math benchmark, CoT prompting improved accuracy from around 18% to over 57% with PaLM 540B. The technique has evolved into variants like tree-of-thought and self-consistency sampling. CoT is now a standard technique used in models like OpenAI's o1 and o3 series, which integrate reasoning steps directly into their architecture.
Explore the Data
Related Terms
Artificial General Intelligence (AGI)
A hypothetical form of AI that can understand, learn, and apply knowledge across any intellectual task at or above human level, rather than being specialized for specific tasks.
AI Alignment
The research field focused on ensuring AI systems behave in accordance with human values and intentions, particularly as systems become more capable.
ChatGPT
OpenAI's conversational AI assistant, launched in November 2022, which catalyzed the current generative AI boom by demonstrating the capabilities of large language models to a mainstream audience.
Fine-Tuning
The process of further training a pre-trained AI model on a specific, smaller dataset to specialize it for a particular task or domain, requiring far less compute than training from scratch.
Foundation Model
A large AI model trained on broad data that can be adapted to a wide range of downstream tasks — examples include GPT-4, Claude, Gemini, and Llama.
Frontier Model
The most capable and advanced AI models at any given time, typically trained with the largest compute budgets and achieving state-of-the-art performance on benchmarks.
AI Economy Pulse
Every Friday: the 3 AI data points that actually matter this week. Free, forever.
Latest: “AI Investment Hits $42B in Q1 2026 — Here's Where It Went”
No spam, ever. Unsubscribe anytime.