Artificial intelligence has rapidly become part of everyday conversations about education, work and society. Yet while much attention is paid to what AI can do, we spend less time thinking about the language we use to describe it.

Words such as intelligence, learning, memory and hallucination are familiar because we normally use them when talking about people. Applied to AI, however, they can suggest capabilities that these technologies do not necessarily possess. The language surrounding AI can therefore influence what we think these systems understand, how much we trust them and where we place responsibility for their outputs.

This is not simply a question of finding technically correct definitions. Language can determine whether people feel able to participate in conversations about AI at all.

What does it mean when AI 'hallucinates'?

Take hallucination, a word now commonly used when a generative AI system produces information that appears plausible but is inaccurate or unsupported.

Outside AI, however, hallucination refers to a very different human experience – when someone seems to see, hear, smell, taste or feel something that does not exist in reality in that moment. During the Better Words for AI project, participants questioned whether borrowing the word to describe AI can give the misleading impression that a system is capable of thinking or perceiving. More broadly, participants frequently reported misunderstanding AI terminology and asked for clear, plain-language explanations with real examples. 

The same question can be asked of other familiar expressions. Does an AI system really learn in the way a person learns? What does memory mean when we use the word about an AI system? And what assumptions do we make when we describe a technology as intelligent?

Questioning these words does not mean that we need to ban them. Instead, it means recognising that the metaphors and terminology surrounding AI are not neutral. They help shape our understanding of the technology.

From better words for AI to OurAIwords

These questions led to Better Words for AI, a project at The Open University that brought students and staff together to explore AI terminology and co-create a more accessible way of talking about it.

The result is OurAIwords, a collaborative, openly available lexicon hosted on OpenLearn Create.

Rather than treating AI terminology as fixed and uncontested, the project approached language as something to explore together. Students and staff discussed the terms they encountered, the assumptions those terms carried and how difficult concepts might be communicated more clearly. The project was designed around student-staff partnership, with participants contributing to identifying problematic terminology and developing the lexicon itself. 

Their discussions also highlighted that people do not all encounter AI language in the same way. The project identified demand for real-world examples and explicit explanations of common misconceptions. It also identified the usefulness of introductory information, which allows people who are less familiar with AI to begin simply and explore concepts in greater depth. 

A lexicon that can evolve

This matters because the language of AI is not standing still. New technologies appear, existing terms acquire new meanings and different communities may understand the same words differently.

OurAIwords is therefore intended as a living educational resource, rather than a final answer to what AI terminology should mean. The original project conceived the resource as dynamic and collaborative, with terminology open to review and refinement as technologies and communities develop.
 
That also means the name is important. OurAIwords reflects the idea that conversations about AI language should not belong exclusively to technologists or AI specialists. Understanding how AI is described, questioning the words used and developing clearer ways of communicating about it are things in which we can all participate.

The aim is not to replace one set of jargon with another. It is to make it easier for people to ask questions such as:

• What does this AI term actually mean?
• Does the word suggest something misleading about the technology?
• Is there a clearer way of explaining the concept?
• Whose perspective might be missing?
• How might the meaning change as AI develops?

Asking those questions is itself part of becoming more critically AI literate.

Explore it and help shape the conversation

Whether you are encountering AI as a student, educator, researcher, through your work or simply in everyday life, the words used to describe it matter.

Explore OurAIwords, question the terminology and join us in shaping clearer, more inclusive and critically informed ways of talking about AI.