In 2025, the OECD released a system of indicators for measuring artificial intelligence capabilities, which includes nine dimensions: language, social interaction, problem-solving, creativity, metacognition and critical thinking, knowledge learning and memory, vision, operation, and robotic intelligence. This framework describes the current global artificial intelligence index and provides a reference for in-depth research and exploration of artificial intelligence. Here, we will describe the five levels of robotic intelligence and the current capabilities of artificial intelligence in this area.
The OECD believes that AI’s robotic intelligence encompasses six dimensions. Four of these dimensions are related to the tasks themselves: the complexity of the task; the level of abstraction in the task definition, which affects the level of understanding required to determine what to do; the complexity of social interaction required to execute the task, and ethical issues, which implicitly provide a set of constraints that influence how the task is performed. The other two dimensions are related to the work context: the complexity of the environment, the level of uncertainty in the environment, and the way agents interact with the environment.
At the peak level, AI performs multiple complex tasks in unstructured environments with a high degree of creativity in goal-setting. They can refine poorly defined task specifications. These robots can adapt to dynamic conditions, learn from experience, and generalize across a wide range of tasks and environments. They demonstrate advanced reasoning capabilities, common-sense reasoning, and highly skilled social intelligence. Robots at this level understand their limitations and can make ethical decisions, refusing to perform tasks that conflict with legal or moral standards. Typical tasks include robots providing home assistance for the disabled, robots making ethical decisions in diverse and dynamic environments, and high-performance autonomous driving.
Level 4 AI robots perform a variety of tasks with differing complexities. They can adapt to dynamic conditions and adjust their behavior based on changing environments. They understand their limitations and use feedback for improvement. This category of tasks involves long-term, complex goals and is context-dependent. Although robots can handle uncertainty and make decisions in uncertain environments, their solutions may not always be as efficient or effective as those found by humans. Typical tasks include cooking robots selecting ingredients based on availability, autonomous wheelchairs navigating obstacles, and automated aerial navigation near airports.
Level 3 AI robots can perform moderately flexible, multi-step tasks in environments with moderate variability. They can work in moderately changing environments and handle tasks involving multiple loosely defined subtasks. These robots can collaborate with humans, adapt to moderate levels of uncertainty, and manage dynamic changes such as variations in lighting, weather, or unknown object types. They can execute tasks using multiple solutions but may struggle in more unpredictable or dynamic environments. Typical tasks include hospital robots handling transportation and cleaning tasks, robots assisting in furniture assembly, and robot photographers autonomously taking pictures based on learned preferences.
Level 2 AI robots perform predefined tasks in semi-structured environments with some variability. They handle low to moderate levels of uncertainty, such as changes in object placement or environmental layout. Tasks typically have clear success metrics, and robots operate with minimal human interaction. They can perform simple, multifunctional tasks but are limited by their inability to handle more complex or unforeseen changes. Typical tasks include medical transport robots, factory material handling robots, and agricultural robots for fruit picking.
The lowest level AI robots perform simple, repetitive tasks in highly structured and controlled environments. They work in static, deterministic environments that are fully understood and predictable. These robots follow pre-specified instructions and lack the ability to make adaptive decisions or handle unforeseen situations. They do not interact with humans and typically cannot manage even minor changes in the environment. Typical tasks include basic automation assembly in factories, robotic vacuum cleaners, and item sorting systems in logistics operations.
Currently, the most advanced systems, such as automated delivery robots and industrial automation systems, perform at approximately level 2 on a scale. These systems perform well in structured environments with predefined tasks. However, they require adaptive decision-making, creativity, and social intelligence to cope with more complex and unpredictable situations. For example, while robots can navigate in pre-mapped environments, they struggle when tasks involve interacting with humans or adapting to unforeseen changes in the environment.