Can a robot be conscious?
A robot can respond personally without settling whether there is anyone experiencing something inside.
The question of AI and consciousness concerns subjective experience: whether there is something it feels like to be the system. This differs from whether a robot can hold a conversation, recognise a name or move its face. For someone choosing an AI companion, the distinction matters. A convincing personality is an experience for the user; it is not in itself an investigation into the system's inner life.
Four things easily confused
Behaviour is what you can observe. Intelligence here means solving tasks. Memory concerns what information the system can store and use. Consciousness concerns subjective experience. Our example: a character can remember your favourite music and respond warmly. This demonstrates a conversation function and use of information. The example does not establish whether the character experiences warmth, longing or desire. [1]
What researchers look for
In 2023, Butlin and colleagues derived indicators from several theories of consciousness, including global workspace and higher-order representations. Their assessment of the systems analysed found no basis for calling them conscious. This is a dated research assessment, not a conclusion about all AI systems in 2026. The indicators offer a way to examine architecture and function; they are not a universal certificate of consciousness. [1]
New indicators are still not proof
A 2026 preprint investigates a particular indicator of higher-order consciousness in large language models. The researchers report signs of belief-guided action and monitoring of their own processing. This is a limited finding about the models studied. It does not prove subjective experience, and the paper does not test the robot products we cover on Sexrobotten. A system's own statement that it is conscious cannot serve as evidence on its own either. [3]
Does the body matter?
Cameras, sensors and motors can give a system more ways to receive information and act. This provides something concrete to investigate. Our view is that a human-like exterior should count neither as evidence of consciousness nor as an explanation of how the system works. A moving face describes mechanics; the question of inner experience requires a different basis. [1]
Uncertainty must be stated clearly
In 2025, Butlin and Lappas proposed principles for responsible research into possible AI consciousness. A key point is to take uncertainty seriously and communicate responsibly. We therefore do not use product terms such as feelings, soul or true love as research findings. They may be marketing or a description of the user's experience. The debate over the possible moral status of future systems remains open. [2]
Our assessment
On a product page, we assess conversation, memory, control and movement as specific functions. A personal response may have value for the user without us knowing whether the system experiences anything. None of the articles discussed documents that a particular commercial sex robot is conscious. Ask for studies of the specific system if a manufacturer makes that claim.