IQ
Also written: Intelligence Quotient, IQ and EQ, Emotional Quotient
IQ is a measure of cognitive abilities like reasoning and problem-solving.
IQ stands for Intelligence Quotient. It's a score from a test designed to measure a person's general mental abilities. These abilities include things like reasoning, problem-solving, and understanding complex ideas.
Think of it like a standardized score for how well someone can learn and apply knowledge. While IQ tests are common for humans, researchers sometimes discuss AI capabilities in relation to human IQ to understand their progress.
Intelligence Quotient (IQ) is a score derived from standardized tests designed to assess human intelligence. These tests typically measure a range of cognitive abilities, including logical reasoning, spatial visualization, verbal comprehension, and processing speed. IQ scores are normalized to a mean of 100 and a standard deviation of 15.
While IQ is a well-established metric for human cognitive capacity, its direct application to AI is complex and often metaphorical. Researchers may use IQ benchmarks as a loose reference point for evaluating AI progress in specific cognitive tasks, but AI systems do not possess general intelligence in the human sense, and their performance on IQ tests can be brittle or achieved through different mechanisms.
The term IQ is appearing as researchers explore ways to benchmark and understand the rapidly advancing cognitive capabilities of AI models. As AI systems like Claude demonstrate more sophisticated reasoning and problem-solving, comparisons to human intelligence metrics like IQ are being made to gauge progress and potential.
- How people use Claude for support, advice, and companionshipAnthropic · Anthropic News ·
- Shaping Model Behavior in GPT-5.1 - Episode 11OpenAI · The OpenAI Podcast ·
- Rana el Kaliouby, Founder of Affectiva, Deputy CEO of Smart Eye, and Emotion AI PioneerMicrosoft · Behind The Tech with Kevin Scott ·
- Anthropic Claude
- Anthropic studies AI's emotional intelligence and its use for support, advice, and companionship, touching on aspects related to cognitive abilities.
- [object Object]
- [object Object]
The concept of IQ was developed in the early 20th century by psychologists like Lewis Terman, building on the work of Alfred Binet and Théodore Simon who created the first practical intelligence test. Source ↗
Ask an AI chatbot a complex logic puzzle or a series of math problems that require multi-step reasoning. Observe how it breaks down the problem and arrives at a solution, noting its ability to follow logical steps and apply learned rules.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
- HeuristicDouble coding agent task completion with docs linksProviding coding agents with a documentation link significantly increases their task completion rate. · 2 min
Questions people ask
- Can AI have an IQ?
- AI systems can perform tasks that are measured by IQ tests, but they do not possess general intelligence or consciousness in the way humans do. Their 'IQ' is task-specific and achieved through different means than human cognition.
- Is IQ the only way to measure AI intelligence?
- No, IQ is just one potential benchmark. Researchers use many other metrics and benchmarks to evaluate AI performance on specific tasks like language understanding, image recognition, and strategic game playing.
First seen on Subvolts Dec 2, 2025. Updated Oct 5, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.