/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Cambridge Dictionary announces “hallucinate” as its 2023 word of the year, citing the word's new meaning in relation to generative AI

Benj Edwards / Ars Technica :

Ars Technica Benj Edwards

Context & Ripple Effects

The term had already become a practical shorthand for a recognizable failure mode: systems can fill gaps in available information with plausible-sounding output, as earlier coverage explained in accounts of ChatGPT filling informational gaps. Cambridge Dictionary’s recognition shows that the issue had moved from specialist AI discussion into general language.

That broader vocabulary matters because model reliability was already becoming measurable and comparable: Vectara’s hallucination evaluation ranked systems by hallucination rates, while researchers were exploring process supervision as a way to reduce the problem.

First-order effects

  • Cambridge Dictionary formalizes a generative-AI-specific sense of “hallucinate,” giving users, developers, and journalists a common label for confidently wrong model output.
  • The recognition makes reliability failures more legible to nontechnical audiences, rather than treating them as an obscure limitation of chatbots.

Second-order effects

  • A shared term makes it easier for AI vendors and evaluators to frame model-quality comparisons around hallucination rates and mitigation methods.
  • Organizations adopting generative AI gain clearer language for setting expectations around verification, uncertainty, and acceptable errors in AI-assisted work.

Third-order effects

  • If this vocabulary remains embedded in public use, trustworthiness will increasingly be treated as a core product attribute alongside model capability, not merely a research concern.
  • The term’s spread may also sharpen debate over whether “hallucination” accurately describes model behavior, particularly where AI outputs are used for exploration rather than factual retrieval.

The trend: Generative AI’s technical failure modes are becoming part of mainstream language, raising the visibility of reliability as models move into everyday use.

Discussion

  • @benedictevans Benedict Evans on threads
    Sometimes I think hallucinations are a red herring - that even if the LLM was 100% accurate, voice and conversation are often the wrong interface
  • @cambridgewords @cambridgewords on x
    The Cambridge Dictionary #WordOfTheYear2023 is ... 🥁 ‘hallucinate’ (verb): When an artificial intelligence hallucinates, it produces false information. 🏆 Find out more about #hallucinate ⬇️ https://dictionary.cambridge.org/ ... #CambridgeWOTY [video]
  • @cambridge_cl @cambridge_cl on x
    If we needed any more reminders of how widespread the influence of AI is, here's a fascinating one from @CambridgeWords Cambridge Dictionary #WordoftheYear2023 ! https://www.cam.ac.uk/...