Cuestionario vinculado a una guía · Medio Nivel

Detectar citas falsas en el cuestionario de IA

Pregúntese por qué los chatbots inventan referencias, los cuatro tipos de malas citas y un método paso a paso para verificar cada fuente.

Rutas de guía relacionadasDetectar citas falsas de IA
Pregunta 1 de 8

¿Por qué un chatbot puede producir una cita que no existe?

Vista previa de cada pregunta8 preguntas

Las respuestas y explicaciones aparecen a medida que realiza el cuestionario.

  1. ¿Por qué un chatbot puede producir una cita que no existe?
    • It copies citations from a secret database of fake papers
    • It predicts likely text, and citations follow a predictable format it can imitate
    • It is programmed to test whether students check sources
    • It only makes up citations when asked about recent events
  2. ¿Qué tipo de mala cita considera la guía la más difícil de detectar?
    • A completely made-up reference
    • A real author paired with an invented title
    • A reference with the wrong year
    • A real, correctly cited paper that doesn't support the claim
  3. ¿Qué pasó en Mata vs. Avianca en 2023?
    • Lawyers were sanctioned for citing cases ChatGPT invented
    • A university banned chatbots in research
    • A publisher sued an AI company over citations
    • A court ruled AI citations are always admissible
  4. Al verificar un DOI, ¿qué debe confirmar después de abrirlo a través de doi.org?
    • That the page loads quickly
    • That the DOI is shorter than 20 characters
    • That the paper it opens has the same title and authors as the citation
    • That the journal is free to read
  5. ¿Qué debe hacer si no puede verificar una fuente?
    • Keep it but move it to a footnote
    • Remove it
    • Ask the chatbot to confirm it
    • Replace the DOI with a URL
  6. Why is asking the chatbot 'Is this citation real?' not a reliable check?
    • Chatbots can't read their own earlier messages
    • It takes too long
    • Chatbots always say the citation is fake
    • The model can produce a confident yes as easily as it produced the fake reference
  7. What kind of error remains common in retrieval-augmented tools, according to the guide?
    • Misattribution, where a real linked source doesn't fully support the claim
    • DOIs that are always missing
    • Citations only in foreign languages
    • Refusing to give any sources
  8. Which of these is a red flag listed in the guide?
    • The paper has more than three authors
    • The source is older than five years
    • A journal name that is slightly off from the real journal
    • The article has an abstract