While, in the traditional sense, the verb “hallucinate” means to see, hear, feel, or smell something that does not exist, another meaning of the word has been increasingly used in a year where AI has fully captured the public’s imagination.

When artificial intelligence produces false information, it “hallucinates.” Although the response may initially appear to be true to those without prior knowledge on the subject, the information given is objectively incorrect.

Accompanying the recently-updated meaning of “hallucinate” in the Cambridge Dictionary are some examples for how it can be used in practice, such as “The latest version of the chatbot is greatly improved but it will still hallucinate facts.”

While its lexicographers have incorporated other AI-related definitions into the dictionary over 2023 — including “generative AI” — the team decided on “hallucinate” as the Word of the Year due to the representative nature of it.

“The new meaning gets to the heart of why people are talking about AI,” they wrote on the Word of the Year announcement page. “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.”

Hallucinations are a significant weakness of AI at present. As an example, and back in February, Google demonstrated the abilities of Bard, its ChatGPT competitor. However, the tool hallucinated facts about the James Webb Space Telescope. After the mistake was noticed, it then led Alphabet, Google’s parent company, to lose around $100 billion in value.

Speaking on “hallucinations” as a term, Dr Henry Shevlin, an AI ethicist and philosopher based at the University of Cambridge, remarked that “It’s striking to me that of all the words we could have collectively decided on to describe the mistakes that large language models make, we went with ‘hallucinate,’ when existing computer-specific terms like ‘glitches’ or ‘bugs’ could have perfectly sufficed.

“Perhaps we chose this vivid psychological verb because it’s so easy to anthropomorphise these systems, treating them as if they have minds of their own. Already, many people think of ChatGPT as closer to a coworker, rather than just another app or utility.”

Recommended reading

• Scot-Secure West | The AI Genie is Out of the Bottle
• ChatGPT Can Once Again Browse the Internet for Up-to-date Information
• OpenAI Faces Lawsuit Over Potential ChatGPT GDPR Violations

Earlier today, DIGIT reported on the announcement that OpenAI, the company behind ChatGPT, has paused new users from subscribing to its paid service following a new surge in demand. The increase in usage comes after OpenAI’s developer conference, DevDay, which was held last week.

At the conference, the company revealed new upgrades and offerings, including “GPTs” — custom versions of the chatbot which users can create themselves for a specific purpose. The tailored versions of ChatGPT, which can be created for aims like generating cooking recipes or helping troubleshoot common tech issues, can also be shared publicly through the upcoming GPT Store.

People who share their GPTs through the GPT Store are also set to earn money from their creations, based on usage.