Nielsen Norman Group

The ELIZA Effect: Why We Love AI

thumbnail

The ELIZA Effect: Why We Love AI

  • Users often attribute human-like characteristics to artificial systems, a phenomenon known as the ELIZA effect.
  • Conversational AI interfaces, such as OpenAI's ChatGPT and Microsoft's Bing Chat, feel personable and human-like, leading to a sense of attachment for some users.
  • Some people have even expressed fears of AI sentience and mistaken AI for true intelligence.
  • However, the fact that chatbots feel human does not necessarily make them good AI products.
  • Companies like Meta are now introducing AI personalities for users to interact with and learn from.
  • ELIZA, a simple chatbot developed in the 1960s, provides a valuable lesson in considering the current AI boom.
  • ELIZA acted as a text-based therapist, repeating user statements and asking questions based on keywords.
  • When ELIZA failed to identify a keyword, it would respond with generic phrases.
  • This example highlights the importance of not conflating human-like attributes with actual technical advancements in AI.