Glossary
What do we mean by that?
A conversation can change when we notice we mean different things by the same word. Here are 21 terms from the book.
- AI alignment
- The effort to make AI systems behave in accordance with intended human goals, values and constraints.
- AI avatar dialogue
- A conversation with an explicitly simulated perspective, such as that of a thinker, mentor or practitioner.
- Automation bias
- Giving automated advice undue weight, including accepting it when relevant evidence calls it into question.
- Co-curating
- Selecting, connecting, arranging and refining material in collaboration with AI.
- Cognitive atrophy
- Weakening of mental abilities through insufficient use or practice.
- Cognitive offloading
- Moving part of a mental task onto an external aid, such as a notebook, calculator or AI assistant.
- Collaborative fluency
- The ability to work effectively with AI while keeping your purpose, judgement and understanding involved.
- CompensAItion
- Deliberately keeping valued thinking abilities in practice while benefiting from AI assistance.
- Dialogical self
- A view of the self as containing different positions that can enter into dialogue with one another.
- Digital scaffolding
- Support from digital tools that helps a learner attempt work they cannot yet manage independently.
- Epistemic vigilance
- The processes and habits used to assess information and the trustworthiness of its source.
- Human-first principle
- Begin with your own purpose, values and criteria when inviting AI into your work.
- Hybrid intelligence
- Combining human and machine capabilities in a shared activity or system.
- Intentional stance
- Explaining or predicting a system’s behaviour by treating it as if it had beliefs, desires and intentions.
- Meta-prompting
- Using AI to help clarify and develop the instructions for a task.
- Metacognition
- Awareness of your own thinking, together with the ability to monitor and regulate it.
- Misalignment
- A mismatch between an AI system’s behaviour or objectives and the human intentions or requirements it should serve.
- Multimodality
- Working or communicating across more than one mode, such as text, images, diagrams and sound.
- One Universal Communication
- The book’s proposition that many habits of effective human communication also help us work with AI.
- Shared mental model
- A sufficiently compatible understanding of a task or situation to support coordination.
- Ventriloquising
- Speaking through another voice; in this context, presenting AI-generated work with little understanding or involvement of your own.