Agi-Friend

An analysis of the requirements for an Artificial Scientist, evaluating logicist, emergentist, and universalist AGI approaches, and proposing a hybrid path forward.

A comprehensive survey exploring the integration of commonsense reasoning into conversational AI systems, covering methods, benchmarks, and current limitations.

This paper presents a domain ontology for AGI containment, identifying core constructs and situating containment strategies within the broader field of cyber science.

Introducing the DICES dataset for nuanced safety evaluation of conversational AI, capturing diverse human perspectives across demographics to move beyond single ground-truth approaches.

Analysis of factors influencing resistance and adoption of AI Voice Assistants using a dual-factor model combining Status Quo Bias and Technology Acceptance Model.

A comprehensive analysis of potentially harmful traits in AI companions, their causal pathways, societal impacts, and design recommendations for risk mitigation.

Analysis of SELMA, a multimodal LLM integrating audio and text for unified virtual assistant tasks like voice trigger detection, device-directed speech detection, and ASR.

Analysis of a Wizard of Oz experiment simulating API usage dialogues to build a training dataset for software engineering virtual assistants.