Glossary
Term | Definition (according to Nigel) |
---|---|
Alignment | The challenge of ensuring the goals and values of AIs match the goals and values of humanity. |
Agent | An AI that can work on a task without human instruction or supervision is said to be an agent or agentic AI. |
AGI | Artificial General Intelligence. An advanced level of AI that can think and reason on a broad range of topics to the same level as an educated adult human. AGIs learn from their experiences and become more intelligent. |
ANI | Artificial Narrow Intelligence. A low level of AI that can only do a specific “narrow” task and cannot learn new skills for itself. Also referred to as weak intelligence. |
Anthropic | AI research company with a focus on AI safety. Creators of Claude. |
ASI | Artificial Superintelligence. An extremely advanced AI that can think, learn, and reason far better than a human. We haven’t invented ASI yet. |
Bias | AIs learn from humans and inevitably pick up... |