---
date: '2023-01-29T10:57:24'
hypothesis-meta:
  created: '2023-01-29T10:57:24.658922+00:00'
  document:
    title:
    - 2301.11305.pdf
  flagged: false
  group: __world__
  hidden: false
  id: tnNraJ_DEe2YBceDAVt0Uw
  links:
    html: https://hypothes.is/a/tnNraJ_DEe2YBceDAVt0Uw
    incontext: https://hyp.is/tnNraJ_DEe2YBceDAVt0Uw/arxiv.org/pdf/2301.11305.pdf
    json: https://hypothes.is/api/annotations/tnNraJ_DEe2YBceDAVt0Uw
  permissions:
    admin:
    - acct:ravenscroftj@hypothes.is
    delete:
    - acct:ravenscroftj@hypothes.is
    read:
    - group:__world__
    update:
    - acct:ravenscroftj@hypothes.is
  tags:
  - chatgpt
  - detecting gpt
  target:
  - selector:
    - end: 22349
      start: 22098
      type: TextPositionSelector
    - exact: "Empirically, we find predictive entropy to be positively cor-related\
        \ with passage fake-ness more often that not; there-fore, this baseline uses\
        \ high average entropy in the model\u2019spredictive distribution as a signal\
        \ that a passage is machine-generated."
      prefix: tropy) predictive distributions.
      suffix: ' While our main focus is on zero'
      type: TextQuoteSelector
    source: https://arxiv.org/pdf/2301.11305.pdf
  text: this makes sense and aligns with the [gltr](http://gltr.io) - humans add more
    entropy to sentences by making unusual choices in vocabulary that a model would
    not.
  updated: '2023-01-29T10:57:24.658922+00:00'
  uri: https://arxiv.org/pdf/2301.11305.pdf
  user: acct:ravenscroftj@hypothes.is
  user_info:
    display_name: James Ravenscroft
in-reply-to: https://arxiv.org/pdf/2301.11305.pdf
tags:
- chatgpt
- detecting gpt
- hypothesis
type: annotation
url: /annotations/2023/01/29/1674989844

---



 <blockquote>Empirically, we find predictive entropy to be positively cor-related with passage fake-ness more often that not; there-fore, this baseline uses high average entropy in the model’spredictive distribution as a signal that a passage is machine-generated.</blockquote>this makes sense and aligns with the [gltr](http://gltr.io) - humans add more entropy to sentences by making unusual choices in vocabulary that a model would not.