In a thought-provoking analysis published on Forbes.com, the topic of subliminal messaging in generative AI and large language models (LLMs) is explored. The article delves into the possibility of subliminal messages being embedded in AI-generated content, either intentionally or inadvertently, and the potential implications of such messaging. While subliminal messaging is not a new concept, its application in the realm of AI raises intriguing questions.
The article highlights the historical background of subliminal messaging, tracing its origins back to ancient Greek studies. It then discusses the use of subliminal messages in movies during the 1940s and 1950s, where frames with messages like “Eat Popcorn” or “You Are Thirsty” were inserted to influence theatergoers’ behavior. The debate surrounding the effectiveness of subliminal messaging is examined, with some dismissing it as mere speculation while others argue for its potential impact on the subconscious mind.
The research on subliminal messaging is also explored, with a focus on whether its effects are short-term or long-term. A neuroscience study cited in the article suggests that subliminal messages can have long-term impacts on decision-making, challenging the notion that their influence is fleeting.
Bringing generative AI into the discussion, the article explains that AI has primarily been used for text-based interactions and image generation. However, the emergence of multi-modal generative AI, incorporating images, video, audio, and text, opens up new possibilities for subliminal messaging. The article outlines four major modes through which generative AI can incorporate subliminal messaging: video-based, image-based, audio-based, and text-based.
The article then delves into the means by which subliminal messages can arise through AI. It suggests that human-instigated subliminal messaging is the most likely path, where AI makers may program generative AI to insert subliminal messages based on sponsorship deals or other motivations. The article also acknowledges the possibility of subliminal messages arising unintentionally through AI’s computational processes, although it cautions against attributing intentionality to AI at this stage.
The ethical implications of subliminal messaging by AI are examined, with contrasting viewpoints presented. While some argue that any form of subliminal messaging is a violation of human autonomy, others consider the potential positive impact if messages promote well-being or positive actions. The article emphasizes the need for careful consideration and regulation of AI-based subliminal messaging to prevent potential harm.
To illustrate the concept of text-based subliminal messaging, the article provides an example using ChatGPT, a popular generative AI app. It demonstrates how the AI can subtly steer the user’s decision-making process through carefully crafted responses, without overtly imposing its influence.
The article concludes by highlighting the ubiquity of generative AI in our daily lives and the potential scale of subliminal messaging’s impact. It raises the question of whether action should be taken now or if society will only address the issue once its manipulation becomes widely recognized. The article ends on a lighter note, sharing a humorous quote by comedian Steven Wright about subliminal advertising.