Generative AI models should include detection mechanisms as a condition for public release
Authors:
Open Access
- 28 October 2023
- journal article
- research article
- Published by Springer Nature in Ethics and Information Technology
- Vol. 25 (4) , 1-7
Abstract
The new wave of ‘foundation models’—general-purpose generative AI models, for production of text (e.g., ChatGPT) or images (e.g., MidJourney)—represent a dramatic advance in the state of the art for AI. But their use also introduces a range of new risks, which has prompted an ongoing conversation about possible regulatory mechanisms. Here we propose a specific principle that should be incorporated into legislation: that any organization developing a foundation model intended for public use must demonstrate a reliable detection mechanism for the content it generates, as a condition of its public release. The detection mechanism should be made publicly available in a tool that allows users to query, for an arbitrary item of content, whether the item was generated (wholly or partly) by the model. In this paper, we argue that this requirement is technically feasible and would play an important role in reducing certain risks from new AI models in many domains. We also outline a number of options for the tool’s design, and summarize a number of points where further input from policymakers and researchers would be required.Keywords
Funding Information
- Victoria University of Wellington
This publication has 14 references indexed in Scilit:
- Ethical Considerations of Using ChatGPT in Health CareJournal of Medical Internet Research, 2023
- Detecting AI content in responses generated by ChatGPT, YouChat, and Chatsonic: The case of five AI content detection toolsJournal of Applied Learning & Teaching, 2023
- Testing the Ability of Teachers and Students to Differentiate between Essays Generated by ChatGPT and High School StudentsHuman Behavior and Emerging Technologies, 2023
- A systematic literature review on the effectiveness of deepfake detection techniquesJournal of Cyber Security Technology, 2023
- Human heuristics for AI-generated language are flawedProceedings of the National Academy of Sciences of the United States of America, 2023
- Survey of Hallucination in Natural Language GenerationACM Computing Surveys, 2023
- Auditing Large Language Models: A Three-Layered ApproachSSRN Electronic Journal, 2023
- Competitive Attention, Superstars and the Long TailSSRN Electronic Journal, 2021
- Labeling Food Processes: The Good, the Bad and the UglyApplied Economic Perspectives and Policy, 2017
- On Best-Possible Obfuscation2007