Zero-shot prompting asks an AI model to perform a task without prior examples or task-specific context. It relies on the model’s general training, making it a simple starting point for everyday questions and requests. section 4.2.1 section 4.2
Use zero-shot prompting for straightforward tasks such as:
- Answering a factual question, such as “Why were cats revered in ancient Egypt?” section 4.2.1
- Making a basic request, such as “Write a short email confirming a meeting.” section 4.2.1
- Requesting general workplace guidance, such as “How can I be a good project manager?” section 5.1.1
A practical workplace prompt
Prompt: “How can I be a good project manager?” The response may offer broad advice about goals, planning, communication, teamwork, risk, quality, budget, adaptability, and time. This is useful immediately, but it may not match a particular project or organization without further refinement. section 5.1.1
Zero-shot output can be factually sound yet misaligned with personal preferences or specific objectives. Refine the prompt when you need a particular format, audience, scope, or working style, and verify the result rather than treating it as automatically correct. section 4.2.1 section 4.10
As models become more capable, they can handle basic tasks with less explicit instruction. Examples still matter when the goal is customization: showing how you want the result differs from merely teaching the task. section 4.2.5
