Prompt Refusal

Data Skeptic

Jul 24 2023 • 44 mins

The creators of large language models impose restrictions on some of the types of requests one might make of them.  LLMs commonly refuse to give advice on committing crimes, producting adult content, or respond with any details about a variety of sensitive subjects.  As with any content filtering system, you have false positives and false negatives.

Today's interview with Max Reuter and William Schulze discusses their paper "I'm Afraid I Can't Do That: Predicting Prompt Refusal in Black-Box Generative Language Models".  In this work, they explore what types of prompts get refused and build a machine learning classifier adept at predicting if a particular prompt will be refused or not.

You Might Like

Acquired
Acquired
Ben Gilbert and David Rosenthal
Darknet Diaries
Darknet Diaries
Jack Rhysider
Hard Fork
Hard Fork
The New York Times
Marketplace Tech
Marketplace Tech
Marketplace
WSJ’s The Future of Everything
WSJ’s The Future of Everything
The Wall Street Journal
Rich On Tech
Rich On Tech
Rich DeMuro
TechStuff
TechStuff
iHeartPodcasts
The Vergecast
The Vergecast
The Verge
Fortnite Emotes
Fortnite Emotes
Lawrence Hopkinson
Waveform: The MKBHD Podcast
Waveform: The MKBHD Podcast
Vox Media Podcast Network