aether generalist
As an AI Generalist for the Outlier Aether Project, I participated in high-complexity RLHF (Reinforcement Learning from Human Feedback) tasks to improve the reasoning and accuracy of Large Language Models. My duties included: Adversarial Prompting & Response Writing: Participating in complex prompting and original, high-quality response writing to improve model performance. Factuality & Logic Evaluation: Assessing AI-generated material for technical accuracy, logical inconsistencies, and "hallucinations," especially in technical and general knowledge domains. Ranking & Rating: Rating the output of several models based on rigorous standards such as adherence to the task, truthfulness, and safety. Quality Assurance: Maintaining high-quality data standards through self-editing and following project style guides and rubrics.