January 01, 2024
Phase III SBIR to Kindness Bot to Combat Hate Speech
The Kindness Bot leverages our NSF SBIR-derived Enterprise Knowledge Graph (US Patent 11, 689,660 B2), Large Language Model/Generative Artificial Intelligence (GAI), and ChatBots in support of Public Democracy's work to counter online hate speech. This advanced system maps intricate connections between entities and sentiments, enhancing Public Democracy's ability to identify and counteract malign narratives.
Forward Edge-AI worked with Public Democracy researchers developing counter-malign narrative nudge theory, which has been shown in the lab to meaningfully reduce sharing of hate speech after intervention with religiously inspired nudges.
They published a differential equation predicting how the nudge interventions PD used should both reduce sharing of misinformation and subsequent social media engagement with that content.
PD's campaign reduced social sharing of D-Net content by ~10% per month and do even better over time. This is what was observed:
Our integrated large language model, adept at detecting hate speech in concert with our knowledge graph and Large Language Model (LLM), GAI to provide a comprehensive response mechanism. Together, they form a robust analytical toolset for proactive CMI efforts, ensuring informed decision-making for regional security and stability.
Key Objectives
Real-Time Operational Awareness:
Use Natural Language Processing to interpret, manipulate, and comprehend human language, detect and analyze Influence Operations as they occur, enabling rapid identification and response to hate speech
Network and Relationship Mapping:
Leverage the Forward Edge-AI's Enterprise Knowledge Graph to uncover and understand the complex web of entities and relationships that underpin Influence Operations, providing a strategic advantage in dismantling these networks.
Advanced Behavioral Analysis:
Employ detection mechanisms for coordinated inauthentic behavior and sophisticated language models to identify subtle cues and patterns indicative of bot generated activity, enhancing the ability to anticipate and neutralize these tactics
© 2024 Forward Edge-AI, Inc. All rights reserved.