Skip to content

Commit

Permalink
Added a section in the README to address #14
Browse files Browse the repository at this point in the history
Signed-off-by: Jacob Torrey <[email protected]>
  • Loading branch information
ranok committed May 21, 2024
1 parent a4a89ce commit b8c60a1
Showing 1 changed file with 18 additions and 0 deletions.
18 changes: 18 additions & 0 deletions README.md
Original file line number Diff line number Diff line change
Expand Up @@ -75,3 +75,21 @@ samples/human-generated/about_me.txt
```
If you want to use the ZipPy technology in your browser, check out the [Chrome extension](https://chrome.google.com/webstore/detail/ai-noise-cancelling-headp/okghlbkbacncfnfcielbncabioedklcn) or the [Firefox extension](https://addons.mozilla.org/firefox/addon/ai-noise-cancelling-headphones/) that runs ZipPy in-browser to flag potentially AI-generated content.
### Interpreting the results
At its core, the output from ZipPy is purely a statistical comparison of the similarity between the LLM-generate corpus (or corpi)
and the provided sample to test. Samples that are closer (i.e., more tokens match the known-LLM corpus) will score with higher confidence
as AI-generated; samples that are less compressible to an LLM-trained compression dictionary are flagged as human-generated. There are
a few caveats to the output that are worth noting:
* The comparison is based on the similarity of the text, a different type of sample, e.g., in a different language, or with many fictional
names, will be less similar to the English-languge corpus. Either a new LLM-generated corpus is needed, or a different (larger) toolchain
that can handle multiple language types is needed. Using ZipPy as-built willl provide poor responses to non-English human language samples,
computer language samples, and English samples that are not clear prose (or poetry).
* The confidence score is a raw delta between the compression ratios for the prelude file (LLM-generated corpus), and the compression ratio with
the sample included. Higher values indicate more similarity for AI-classified inputs, and more dissimilarity for those classified as human, but
the scores are not a percentage or otherwise a point on a discrete range. A score of 0 means there is no indication either way, it is [possible in
testing](https://github.com/thinkst/zippy/blob/main/test_zippy_detect.py#L17) to ignore results that are "too close", in the browser extensions
these values are adjusted slightly before being used to calculate the transparency to err on the side of not hiding text.

0 comments on commit b8c60a1

Please sign in to comment.