The Cambridge Dictionary
Word of the Year 2023 is…
verb
When an artificial intelligence hallucinates, it produces false information.
The potential of generative artificial intelligence continues to be a lively topic of discussion. The surge of interest this year has been prompted by an abundance of tools being released for public use, such as ChatGPT, Bard, DALL-E, and BingAI, all of which depend on large language models (complex mathematical representations of language that are based on very large amounts of data and allow computers to produce language that seems similar to what a human might say).
With this new technology comes new language. Throughout 2023, the Cambridge Dictionary lexicographers have added new AI-related definitions to the dictionary, including large language model (or LLM), generative AI (or GenAI), and GPT. Many existing words, such as hallucinate, have adopted additional AI-related meanings this year as well.
The Cambridge Dictionary team chose hallucinate as its Word of the Year 2023 as it recognized that the new meaning gets to the heart of why people are talking about AI. Generative AI is a powerful tool but one we’re all still learning how to interact with safely and effectively – this means being aware of both its potential strengths and its current weaknesses.
Why hallucinate?
What does it say about 2023?
“At Cambridge, we’re in the business of providing authoritative information about language for learners of English, so we’re excited by the potential of AI-enabled access to our content. Managing the tendency of generative AI tools to hallucinate will be key to ensuring our users can continue to trust us.
Wendalyn Nichols, Publishing Manager
Cambridge Dictionary
Take a deeper look at hallucinate as a word
Find out more
Hear from Dr Henry Shevlin, AI expert
Check out these past winners of the Cambridge Dictionary Word of the Year.
Previous Word of the Year winners
Generative AI has sparked a huge amount of worldwide discussion this year due to its potential positive impact across healthcare, transportation, finance, and education.
2016 paranoid
hallucinate
Update and add in 2022 – homer
Although generative AI has gained a lot of traction this year in particular, people around the world have actually been using AI for decades. Products and services that offer personalization, whether it be recommending a TV show to watch or a music playlist that you’ll enjoy, are all made possible by AI. Digital assistants and smart home devices, such as Apple products that use Siri or Amazon Alexa devices, use natural language processing technology (the branch of computer science that involves giving computers the ability to interpret and produce language) to deliver the desired outcomes.
The everyday use of AI tools may be familiar to all, but the explosion in the use of generative AI tools is a new phenomenon. In 2023 it became clear that this newly available technology is far from perfect as it’s capable of producing false information – hallucinations – and presenting this information as fact.
As excitement meets reality, the limits of generative AI tools are as fascinating as their potential. AI hallucinations remind us that humans still need to bring their critical thinking skills to the use of these tools. Large language models are only as reliable as the information their algorithms learn from. Human expertise is arguably more important than ever, to create the authoritative and up-to-date information that LLMs can be trained on.
Dictionary or AI?
As AI tools become more powerful, and are used more and more in our daily lives, you might be tempted to ask a chatbot for help the next time you need to know the meaning of a word. It’s true that generative AI has an impressive ability to answer questions on a wide variety of topics. If you are a good prompt engineer, you might even be able to ask an AI to write a dictionary definition for you.
This is exciting news for the future, but – at least for now – the Cambridge Dictionary team believes that traditional dictionaries are still the best way to improve your knowledge of English vocabulary.
AI vs Cambridge Dictionary
Researchers have not found a way to prevent AIs from hallucinating, which means that you can’t be sure that the information they give you is accurate.
To get the best results out of AI tools, you need to write detailed prompts and adjust them until you get what you’re looking for – and the results can be long and hard to read.
Large language models often produce invented example sentences that are grammatically incorrect or unnatural.
AI tools don’t necessarily give relevant explanations or usage tips for learners of English.
Generative AI simply responds to your prompts.
All definitions are written and checked by experts. The Cambridge team has many years of English language teaching experience, and we improve our dictionary based on the latest research.
Getting results is easy! Just type a word in the search field to get exactly what you’re looking for. All our English definitions are written with learners in mind. We also offer 28 bilingual dictionaries, and you can navigate the site in any one of 18 different languages.
All example sentences are based on authentic evidence of how real English speakers speak and write. They are carefully chosen to reflect important usage patterns.
We offer so much more than just definitions. Log in to Cambridge Dictionary +Plus to build your vocabulary with word lists and quizzes created by Cambridge experts. Create your own word lists and quizzes to share with our dictionary community.
Our resources are designed especially to help learners gain a comprehensive understanding of words and how to use them. For example, the Thesaurus explains how similar words differ in meaning and helps you choose the best one. Grammar tips offer useful information about common learner errors.
Our Word of the Year word list
Expand your vocabulary and test yourself with our Words related to artificial intelligence word list, specially created by the Cambridge Dictionary team. You can find this word list and many more by creating a free account on the Cambridge Dictionary website.
Take the quiz!
Read our Word of the Year blog post to discover more AI terms that have recently come into the English language.
Learn about words on our blog
New words about AI
The field of AI is growing very quickly, and this means that lots of neologisms (new words) related to AI are entering the English language. Some of these words will be ephemeral, or soon forgotten, but it is likely that others will go on to be used more widely. The Cambridge Dictionary team monitors these trends closely. When our lexicographers have collected enough evidence to suggest that a new word is here to stay, they add it to the dictionary.
Discover some of the words about AI that the Cambridge Dictionary team is considering. Find out what they mean and vote on whether you think they should be added to the dictionary!
Vote now
ghost work
RLHF
voice
cloning
bias
super-intelligent
train
Exploring the language of AI
Do you know the other meaning of hallucinate? Find out what it is, learn more about how a new sense of hallucinate was identified and added to the dictionary, and explore other ways in which artificial intelligence is making its mark on the vocabulary of English.
What are hallucinations and what more can we expect from AI?
Here's why:
AI products provide computer-generated audio or no audio at all.
Pronunciation audio features real speakers of British and American English.
Read the blog
The emergence of a new meaning of hallucinate is a great case in point. It’s human experts tracking and capturing changes in the language that make the Cambridge Dictionary a trustworthy source of information about new words and senses – ones the public-facing AI tools won’t have learned yet.”
How many AI-related terms do you know?
wiggle animation
Exploring the language of AI
Back to previous page
New evidence from the corpus showed that the subject of hallucinate can also be an AI. This change in subject was one of the clues that showed our researchers that a new meaning of hallucinate had developed which was specific to AI.
Cambridge lexicographers continually look at examples of how words are used in a wide range of written and spoken English sources to help them identify patterns of usage like this. They then use real-world knowledge and experience to craft definitions that accurately explain new words and senses.
What did we find out about hallucinate ?
In the past, the subject of the verb hallucinate was typically a human being. When someone hallucinates, they seem to hear, feel, or smell something that does not exist.
When an artificial intelligence (= a computer system that has some of the qualities that the human brain has, such as the ability to produce language in a way that seems human) hallucinates, it produces false information.
The corpus also shows us changes in the use of the word AI itself. In the past, people mostly talked about AI as the use or study of computer systems that have some of the qualities that the human brain has. The word behaved in a similar way to the names of other fields of study, such as chemistry or computer science. It was usually uncountable, which means that it did not have a plural form.
As AI has developed, however, it has become more likely that we will encounter particular computer systems that have these human-like properties. We can describe these kinds of systems as AIs, using the word as a countable noun, and so Cambridge lexicographers updated the entry for AI to include this countable sense. We talk about these AIs not as a field of study but as individuals that we can interact with.
The words that collocate with (are used with) AI give us more information about how we talk about this topic. People imagine the future of AI, and ask whether AIs will become super-intelligent – and, if they do, whether they will be friendly or go rogue. At least for the moment, super-intelligent AIs that can truly be said to think like human beings are only found in fiction; nevertheless, a lot of the words used today to talk about the abilities of AIs were originally used to talk about human intelligence and ways of thinking. In addition to hallucinating, AIs can be trained and can show bias. Instead of talking about AI hallucination, some people prefer the word confabulation, which is also a technical term used in psychology. All these words have specialized meanings when they are used about AIs, which you can find defined in the Cambridge Dictionary.
The rapid spread of these new words and senses shows how English speakers use language in innovative and creative ways to understand and talk about new technological developments.
super-intelligent
Our corpus research also helps us to identify existing words that are being used in new ways, and subsequently require additional meanings. We are always collecting new examples of English in use to add to our corpus, so you can stay up to date with how the language is used. Hallucinate is a good example of how an existing word can develop a new meaning.
The words that collocate with (are used with) AI give us more information about how we talk about this topic. People imagine the future of AI, and ask whether AIs will become super-intelligent – and, if they do, whether they will be friendly or go rogue. At least for the moment, super-intelligent AIs that can truly be said to think like human beings are only found in fiction; nevertheless, a lot of the words used today to talk about the abilities of AIs were originally used to talk about human intelligence and ways of thinking. In addition to hallucinating, AIs can be trained and can show bias. Instead of talking about AI hallucination, some people prefer the word confabulation, which is also a technical term used in psychology. All these words have specialized meanings when they are used about AIs, which you can find in the Cambridge Dictionary.
The words that collocate with (are used with) AI give us more information about how we talk about this topic. People imagine the future of AI, and ask whether AIs will become super-intelligent – and, if they do, whether they will be friendly or go rogue. At least for the moment, super-intelligent AIs that can truly be said to "think" like human beings are only found in fiction; nevertheless, a lot of the words used today to talk about the abilities of AIs were originally used to talk about human intelligence and ways of thinking. In addition to hallucinating, AIs can be trained and can show bias. Instead of talking about AI hallucination, some people prefer the word confabulation, which is also a technical term used in psychology. All these words have specialized meanings when they are used about AIs, which you can find in the Cambridge Dictionary.
go rogue
confabulation
and analysed by our lexicographers
to inform new and updated entries in the Cambridge Dictionary.
Examples of English in use are added to the Corpus
Large language models, the complex mathematical representations of language that allow AIs to respond to prompts in ways that seem human, are developed by training them on huge quantities of data, for example written texts. When we add words to the Cambridge Dictionary, we also consult large collections of language data – in our case, the Cambridge English Corpus. We use the corpus to find evidence for new English words and how they are used. Our team of lexicographers interprets this evidence using their expert knowledge of English and their teaching experience, bringing together data and human insight to make brilliant dictionary entries.
For example, our lexicographers consciously built a definition of hallucinate that would make it clear to readers that AIs do not have feelings or thought processes like human beings do:
We can’t know for certain how AI will change the way we live and work in the future, but our Word of the Year shows that it has already changed how we talk.