Google's Chief Scientist, Jeff Dean, reportedly avoids in-depth conversations about Artificial General Intelligence (AGI) due to the term's ambiguous and varied definitions. Dean notes that the perceived difficulty of achieving AGI differs wildly depending on the definition used. He acknowledges that current AI models already surpass the average human in many non-physical tasks. However, these models still fall short of expert-level performance and struggle with tasks requiring specialised knowledge.
Dean anticipates AI's increasing role in accelerating progress across scientific and engineering domains. He suggests that automated search and computation will drive advancements over the next two decades. Google is extending large-scale language models to modalities beyond text, such as images, video, and audio. This aims to enable understanding of data from diverse sources, including self-driving cars and genomic information. Dean envisions future AI capable of solving complex problems by dividing them into numerous subtasks and executing them with high accuracy.
