What is it?
The copilot metaphor compares the use of an AI tool to assist in completing tasks, to working with another pilot in the plane. In aviation, there are commonly two fully trained pilots (co-pilots), who take responsibility for different actions to fly the plane safely to its destination.
What does this metaphor suggest about AI?
Using the copilot metaphor suggests that the AI tool is the equivalent of a trained and capable individual who handles some of the tasks so that you can focus on your own. The term implies that AI is able to understand, make decisions, and complete the tasks without human input. In contrast, using the autopilot metaphor compares the AI system to a computer system that is able to run in the background and make small adjustments when things are running normally.
Where does the metaphor commonly appear?
In addition to the GitHub Copilot, Microsoft picked up the copilot metaphor to name their own AI companion. It is now both a brand name, as well as a metaphor used to describe other tools in marketing and journalism.
What does this metaphor highlight?
The term copilot promotes the idea of AI as being able to take on the role of a trusted and skilled individual with whom you share tasks and responsibilities. There may also be a suggestion that the AI tool is a ‘junior partner’ over which you retain command. Autopilot implies the AI system is able to run regular tasks in the background with minimal intervention from the user.
What does this metaphor hide?
While an AI system can perform tasks and respond to commands, it is not able to ‘step in’ to identify and respond to new challenges with the same knowledge and experience of a skilled individual.
What makes (or could make) this metaphor work?
Aviation metaphors and processes are used in different sectors and in common speech (e.g. “put your own oxygen mask on first” to refer to making yourself safe before you help others). These metaphors either relate to the novelty of human flight, the social changes linked to commercial routes, or to interest and trust in aviation safety.
The copilot metaphor relies on different cultural understanding of these roles. For example, people can think of a copilot as being both the second in command trained pilot (in which the user is in charge), and a partner (in which the user is equal to the AI system).
Whose interests does it tend to serve?
Those writing about AI find different interpretations of the copilot metaphor to be a useful way to communicate about the roles and limitations of AI. For example, this allows people to compare level of intelligence, independence, decision-making, and how the AI system works with the user (the ‘captain’). However, interpreting this as a ‘captain in charge’ and a junior ‘copilot’, can push the responsibility for anything that goes wrong to the user. When these metaphors are used without explanation, it allows people to avoid difficult conversations around the abilities of these tools and responsibilities of those developing them.
