GenAI Mirage: The Impostor Bias and the Deepfake Detection Challenge in the Era of Artificial Illusions

View on arXiv ← Back to list

Authors: Mirko Casu, Luca Guarnera, Pasquale Caponnetto, Sebastiano Battiato

Published: 2023-12-24 10:01:40+00:00

AI Summary

This paper introduces the novel "Impostor Bias," a cognitive bias where individuals question the authenticity of multimedia content due to the prevalence of AI-generated content, leading to erroneous judgments. The paper explores the causes and consequences of this bias and suggests strategies for mitigation.

Abstract

This paper examines the impact of cognitive biases on decision-making in forensics and digital forensics, exploring biases such as confirmation bias, anchoring bias, and hindsight bias. It assesses existing methods to mitigate biases and improve decision-making, introducing the novel Impostor Bias, which arises as a systematic tendency to question the authenticity of multimedia content, such as audio, images, and videos, often assuming they are generated by AI tools. This bias goes beyond evaluators' knowledge levels, as it can lead to erroneous judgments and false accusations, undermining the reliability and credibility of forensic evidence. Impostor Bias stems from an a priori assumption rather than an objective content assessment, and its impact is expected to grow with the increasing realism of AI-generated multimedia products. The paper discusses the potential causes and consequences of Impostor Bias, suggesting strategies for prevention and counteraction. By addressing these topics, this paper aims to provide valuable insights, enhance the objectivity and validity of forensic investigations, and offer recommendations for future research and practical applications to ensure the integrity and reliability of forensic practices.

Key findings

The paper highlights the significant impact of cognitive biases, particularly the newly defined Impostor Bias, on the objectivity of forensic investigations. It reviews existing deepfake detection methods and their limitations, emphasizing the need for more robust and generalized techniques. The authors propose strategies to mitigate biases and improve the reliability of forensic evidence in the age of AI-generated media.

Approach

The paper analyzes cognitive biases in forensic science, specifically focusing on the emerging Impostor Bias related to AI-generated multimedia. It examines existing deepfake detection methods and proposes strategies for mitigating biases in forensic investigations, including training and the use of objective procedures.

Datasets

CelebA, FFHQ, ImageNet, FaceForensics++, UADFV, MSCOCO, Flickr30k

Model(s)

ResNET-18, ResNET-50, Vision Transformers, various GANs (AttGAN, CycleGAN, GDWCT, IMLE, ProGAN, StarGAN, StarGAN-v2, StyleGAN, StyleGAN2), various Diffusion Models (DALL·E 2, GLIDE, Latent Diffusion, Stable Diffusion)

Author countries

Italy

← Previous