US20240412029
2024-12-12
Physics
G06N3/006
Generative AI systems can significantly enhance user interactions by incorporating emotional context into prompt designs. By collecting audio, video, physiological, cognitive, and environmental data from users through sensors, AI can infer the user's emotional state. This state is then used to augment the original prompt, allowing the AI to become more context-aware and responsive to non-verbal cues that users might not explicitly express. The integration of such cues enables AI to interact more naturally with humans, mimicking the nuanced communication seen in human-to-human interactions.
Human communication is a complex mix of verbal and non-verbal signals. While verbal communication involves spoken or written words, non-verbal cues include facial expressions, body language, vocal tone, and more. These cues provide vital context about a person's emotional state and can often convey more meaning than words alone. Studies indicate that in some scenarios, over 90% of communication's meaning comes from non-verbal signals. When verbal and non-verbal messages conflict, people tend to trust the non-verbal signals more.
Traditional large language models (LLMs) are limited in their ability to interpret non-verbal cues since they rely solely on text-based prompts. This limitation means users must accurately translate their intended meaning into text, which can be challenging when trying to convey emotions or contextual nuances. Consequently, LLMs may misinterpret prompts, leading to irrelevant or inappropriate responses. The absence of non-verbal input results in a loss of crucial information necessary for understanding the full context of user interactions.
The proposed approach addresses these limitations by augmenting user prompts with non-verbal cues automatically detected by sensors. By integrating physiological, cognitive, audio, and video signals into prompt design, AI systems can better understand the user's state and intent. This method allows for more intuitive and empathetic AI interactions, as the responses generated are more aligned with the user's emotional context. As a result, users experience higher quality and more satisfying interactions with AI systems.
The automatic augmentation of prompts using emotional context improves communication quality between humans and AI. By sensing non-verbal cues and incorporating them into responses, generative AI systems can offer more relevant and appropriate feedback. This enhancement leads to a more natural and human-like interaction experience for users, reducing misunderstandings and increasing user satisfaction with AI tools.