Researchers at the Beijing Institute for General Artificial Intelligence (BIGAI) argue that Artificial Intelligence (AI) lacks social skills, making it difficult for AI to understand social contexts. According to Lifeng Fan, the first author of the study, AI has transformed our lives and society. However, the next significant challenge for AI is Artificial Social Intelligence (ASI). ASI is a multi-faceted field that includes social perception, Theory of Mind, and social interaction.
Fan believes that ASI is challenging due to its high context-dependency. An ASI system needs to interpret latent social cues, understand other agents’ mental states, and cooperate in a shared task. This requires a comprehensive approach that considers social intelligence’s three essential and inextricably linked aspects: social perception, Theory of Mind, and social interaction.
To develop an ASI system with human-like characteristics, researchers recommend taking a holistic approach that mimics how humans interact with the world around them. This includes using different learning methods such as lifelong learning, multi-task learning, one-/few-shot learning, and meta-learning. They also suggest defining new problems, creating new environments and datasets, setting up new evaluation protocols, and building new computational models. The goal is to equip AI with high-level ASI and improve human well-being with the help of Artificial Social Intelligence.
In-Article Image CreditsNeural network artificial intelligence Forest of synthetic pyramidal dendrites via Wikimedia Commons by Hermann Cuntz with usage type - Creative Commons License. October 3, 2011
Artificial intelligence - Edge detection applied to a photograph via Wikimedia Commons by Jon McLoone with usage type - Creative Commons License. June 8, 2010
Featured Image CreditArtificial intelligence - Edge detection applied to a photograph via Wikimedia Commons by Jon McLoone with usage type - Creative Commons License. June 8, 2010