Artificial general intelligence
What does artificial general intelligence mean and what does it mean for AI research? Find out more here.
Definition
Characteristics and state of research
AGI is still theoretical and has not yet been achieved in practice. An AGI system would need to have general knowledge of the world, contextual learning and transfer learning in order to apply knowledge from one area to another. It would have to solve problems creatively, adapt to new situations and even exhibit characteristics such as consciousness or self-reflection – abilities that current AI systems do not possess. General AI is seen as the ultimate goal of AI research [1]. However, there is currently no system that demonstrates human-like general intelligence; even the most advanced AI (such as large language models) are specialized and have limitations.
Challenges and significance
The development of AGI raises many technical and ethical questions. Technically, it is unclear how all cognitive abilities can be replicated and how much computing power or data would be required. Ethically, there are questions about control and security: an AGI could have an immense impact on society and the economy. Researchers are therefore discussing both possible architectures and guidelines for designing a future strong AI safely and for the benefit of humanity.