Text-to-Image Models Prompted on Intellect

Text-to-image models, like Midjourney and DALL-E, have been shown to reinforce harmful biases, often perpetuating outdated and discriminatory stereotypes. In this study, we delve into a particularly insidious bias largely overlooked in generative image research: Brilliance Bias. By age six, many children begin to internalize the damaging notion that intellectual brilliance is a male trait—a belief that persists into adulthood. Our findings demonstrate that popular image AI models possess this bias, further entrenching the misguided notion that exceptional intelligence is inherently male. This study calls for addressing these biases in AI to ensure a more realistic representation of intellectual capabilities, helping shape a future where talent and brilliance are more broadly recognized.

诸如Midjourney与DALL-E这类文本生成图像模型（Text-to-image model），已被证实会加剧有害偏见，时常固化过时且带有歧视性的刻板印象。本研究深入探究了生成式图像研究领域中长期被大幅忽视的一类极具隐蔽性的偏见：才华偏见（Brilliance Bias）。研究表明，儿童在年满六岁左右便会开始内化这一有害认知：智力才华属于男性特质，这一观念会持续延续至成年阶段。我们的研究结果显示，主流图像人工智能模型均存在此类偏见，进一步强化了"卓越智力本质上属于男性"这一错误观念。本研究呼吁针对人工智能中的此类偏见开展应对与治理，以确保对智力能力的呈现更为客观真实，助力构建一个能够更广泛认可才华与智慧的未来。