System 2
A concept derived from Daniel Kahneman's dual-system theory, referring to slow, deliberate thinking that requires focus and reasoning. In the AI field it corresponds to reasoning models that generate a chain of thought before answering.
Timeline (last 90 days)
System 1/System 2 theory originates from Daniel Kahneman's Thinking, Fast and Slow: System 1 is fast, automatic, intuitive judgment, and System 2 is slow, deliberate reasoning that requires working memory.
JEV-27B-VL exposes two distinct interfaces: System 1 reads logits and returns typed decisions and calibrated probabilities, supporting yes/no judgments, selection among 2 to 256 options, and scoring from 0 to 5, with prompts up to 256K tokens; System 2 is described as unmodified Qwen3.8-27B, which accepts image input and outputs text or step-by-step reasoning.
There is a key disagreement over whether LLMs can exhibit System 2 capabilities: one side holds that System 2 capabilities are already implicit in the weights and only need appropriate prompting or compute budget to be elicited, while the other side holds that what the model learns is merely the surface form of reasoning rather than a transferable reasoning core; there is currently no conclusion.
For tasks requiring multi-step reasoning, building a System 2 scaffold for a System 1 model at the application layer (providing scratch space, introducing external tools for verification, setting checkpoints, allowing multiple attempts and picking the best) is often more effective than simply switching to a larger model.
The recommended collaborative system design is: use a System 1 one-glance model to answer all questions while retaining confidence scores, and hand off tasks below a threshold to a System 2 thinking model.
System 2 is the same layer stack running repeatedly in a loop, writing down the result of each step and reading it back in the next round—this is the essence of chain of thought.
The practical criterion for choosing a model is to count how many steps the answer requires: use System 1 for one-step answers and System 2 for multi-step answers.
System 2 models with recurrent reasoning capability can maintain performance as chains grow longer, because they can explicitly record intermediate conclusions at each step.
A System 2 model is essentially the same network running repeatedly in a loop, forming a thought process through a 'write notes–read notes–advance' scratchpad mechanism.
Reasoning models have introduced mechanisms such as chain-of-thought, reflecting the evolution of AI from System 1 fast intuitive responses toward System 2 slow deep thinking.
All Facts (10)
The practical criterion for choosing a model is to count how many steps the answer requires: use System 1 for one-step answers and System 2 for multi-step answers.
60%UnverifiedSystem 1/System 2 theory originates from Daniel Kahneman's Thinking, Fast and Slow: System 1 is fast, automatic, intuitive judgment, and System 2 is slow, deliberate reasoning that requires working memory.
50%UnverifiedJEV-27B-VL exposes two distinct interfaces: System 1 reads logits and returns typed decisions and calibrated probabilities, supporting yes/no judgments, selection among 2 to 256 options, and scoring from 0 to 5, with prompts up to 256K tokens; System 2 is described as unmodified Qwen3.8-27B, which accepts image input and outputs text or step-by-step reasoning.
50%UnverifiedFor tasks requiring multi-step reasoning, building a System 2 scaffold for a System 1 model at the application layer (providing scratch space, introducing external tools for verification, setting checkpoints, allowing multiple attempts and picking the best) is often more effective than simply switching to a larger model.
50%UnverifiedThere is a key disagreement over whether LLMs can exhibit System 2 capabilities: one side holds that System 2 capabilities are already implicit in the weights and only need appropriate prompting or compute budget to be elicited, while the other side holds that what the model learns is merely the surface form of reasoning rather than a transferable reasoning core; there is currently no conclusion.
50%UnverifiedThe recommended collaborative system design is: use a System 1 one-glance model to answer all questions while retaining confidence scores, and hand off tasks below a threshold to a System 2 thinking model.
50%UnverifiedSystem 2 is the same layer stack running repeatedly in a loop, writing down the result of each step and reading it back in the next round—this is the essence of chain of thought.
50%UnverifiedSystem 2 models with recurrent reasoning capability can maintain performance as chains grow longer, because they can explicitly record intermediate conclusions at each step.
50%UnverifiedA System 2 model is essentially the same network running repeatedly in a loop, forming a thought process through a 'write notes–read notes–advance' scratchpad mechanism.
50%UnverifiedReasoning models have introduced mechanisms such as chain-of-thought, reflecting the evolution of AI from System 1 fast intuitive responses toward System 2 slow deep thinking.
50%