ONLINEAGENT_OPS 2026.Q3 HOME HUBSMODELSBUILDINGCOMPANIESFILMMAKINGSECURITYLIVING WITH ITSKILLSALL HUBS, ONE TERMINAL → ARTICLES CRAFT RECORD BLOG TOOLS SKILLS SEARCH /
HOME›AI MEANINGS›WHAT IS ZERO-SHOT PROMPTING?
AI MEANINGS · MEANINGINTERMEDIATE

What Is Zero-Shot Prompting? Zero-Shot Meaning in Plain English

Zero-shot prompting means asking AI to do a task with instructions only, no examples. When it works, when to add examples, and where the word comes from.

READ5 min
WORDS1,087
SECTIONS7
TYPEMEANING
CHECKED12 OCT 26

Zero-shot prompting means asking an AI to do a task with instructions only, without showing it any examples of the answer you want. Most everyday prompts are zero-shot: you describe the job, and the AI relies on what it already learned to work out the rest.

◈ IN PLAIN TERMS

Imagine asking a new helper to “sort these letters into urgent and not urgent”, and leaving it there. That is zero-shot. Hand them two letters you have already sorted, as a guide, and it becomes few-shot. A “shot” is simply one example, so zero-shot means zero examples.

In a sentence: “I tried it zero-shot first, just asking for a summary, and only added a sample when the tone came out too formal.”

What does zero-shot prompting mean?

The plain point: instructions, no worked examples.

Google’s glossary for its AI tools defines it in one line: zero-shot prompting is to “provide a prompt without any examples and rely on the model’s pre-existing knowledge.” The same glossary calls one example one-shot, and several examples few-shot.1 IBM puts it the same way: the model is not given examples of the output, and leans on its training to infer a good response.2

A simple example

The plain point: a clear instruction is enough for many jobs.

ZERO-SHOT — ADAPT THE BRACKETS
Read the email below and label it URGENT or NOT URGENT. Urgent means someone needs an answer from me today. Reply with the label and one short reason. Email: [paste the email]

There are no sample emails with labels in it, so this is zero-shot. If the labels keep coming out wrong, add two or three labelled examples and it becomes few-shot prompting.

Why it matters to you

The plain point: start without examples, and add them only when the answer misses the format or tone you wanted.

  • It is the normal starting point. IBM says zero-shot is often the default strategy for a new problem.3
  • Reasoning models often need nothing more. OpenAI tells people using its reasoning models to “Try zero shot first, then few shot if needed”.4
  • Not every vendor agrees. Google’s Gemini guide goes the other way: it recommends always including examples and says prompts without them are likely to be less effective.5 That disagreement is why the practical rule is to test: prompting myths sets the vendors side by side.
TAKEAWAY

Ask plainly first. If the answer is right but the shape is wrong, show an example rather than writing a longer description.

Often confused with

  • Zero-shot learning: the older machine-learning idea of recognising things a model never saw labelled examples of during training. IBM describes it as categorising objects or concepts “without having seen any examples of those categories or concepts beforehand.”6 Zero-shot prompting borrows the name; it is about the examples in your message, not the training.
  • Few-shot prompting: the same request with a handful of worked examples added. See few-shot prompting.
  • Fine-tuning: retraining a model on many examples so it changes for good. A zero-shot prompt changes nothing about the model. See fine-tuning.

Where the word comes from

“Zero-shot” started in machine learning before chatbots. A 2009 paper by Palatucci, Pomerleau, Hinton and Mitchell set out “the problem of zero-shot learning”, predicting classes left out of the training data, and pointed to earlier work on “zero-data learning” by Larochelle and colleagues in 2008.7 The prompting sense was set out in OpenAI’s 2020 GPT-3 paper, which tested the model in zero-shot, one-shot and few-shot settings and described zero-shot as giving the model only an instruction, with no demonstrations.8

Questions people ask

What is the difference between zero-shot and few-shot prompting?

Zero-shot gives instructions only; few-shot adds a few examples of the input and the answer you want. Google’s Gemini guide says examples are often used to control the formatting, phrasing and scope of the answer.9

Is zero-shot prompting bad?

No. It is how most people use chatbots, and it works well for clear, common tasks. IBM notes its results vary more on specialised or highly specific tasks, which is where examples help.3

What is zero-shot chain of thought?

Adding a line such as “Let’s think step by step” to a zero-shot prompt so the AI writes out its reasoning. A 2022 paper by Kojima and colleagues named it, and found it helped on reasoning tests without any examples.10 On newer reasoning models it is often unnecessary; see chain of thought.

Learn more on this site

QUICK CHECK · 3 QUESTIONS

Pick an answer to see if you got it. Nothing is recorded.

1. What is zero-shot prompting?

  • Asking with instructions only, and no examples
  • Giving the AI a few worked examples first
  • Retraining the model on many examples
Show the answer

Answer: Asking with instructions only, and no examples. The page says a ‘shot’ is one example, so zero-shot means instructions with zero examples; adding a few is few-shot.

2. The answer is right, but its shape is wrong. What does the page suggest?

  • Write a much longer description
  • Show an example of the shape you want
  • Ask the same question again, unchanged
Show the answer

Answer: Show an example of the shape you want. The page’s takeaway says to ask plainly first, and if the shape is wrong, show an example rather than writing a longer description.

3. Do all AI companies agree you should start zero-shot?

  • Yes, they all say never to use examples
  • Yes, they all say zero-shot is always best
  • No, Google’s guide recommends always adding examples
Show the answer

Answer: No, Google’s guide recommends always adding examples. The page says OpenAI suggests trying zero-shot first, while Google’s Gemini guide recommends always including examples, so the practical rule is to test.

Related terms: AI literacy · AI memory · AI prompt generator · Sycophancy · System prompt · Temperature

SOURCES
  1. 1Google Cloud, Generative AI glossary, last updated 9 Oct 2026, read at source 12 Oct 2026“Zero-shot prompting: provide a prompt without any examples and rely on the model’s pre-existing knowledge.” … “One-shot prompting: provide a single example in the prompt to guide the model’s response.” ↑
  2. 2IBM, What is zero-shot prompting?, undated, read at source 12 Oct 2026“Zero-shot prompting is a prompt engineering method that relies on the pretraining of a large language model (LLM) to infer an appropriate response. In contrast to other prompt engineering methods such as few-shot prompting, models aren’t provided examples of output when prompting with the zero-shot technique.” ↑
  3. 3IBM, What is zero-shot prompting?, read at source 12 Oct 2026“A zero-shot prompting strategy is often the default prompt engineering strategy when tasking a model with a new problem.” ↑a“While zero-shot prompting can be effective, its performance can vary significantly depending on the complexity and specificity of the task.” ↑b
  4. 4OpenAI, Reasoning best practices, undated guide, read at source 12 Oct 2026“Try zero shot first, then few shot if needed: Reasoning models often don’t need few-shot examples to produce good results, so try to write prompts without examples first.” ↑
  5. 5Google, Prompt design strategies, Gemini API docs, last updated 17 Sep 2026, read at source 12 Oct 2026“We recommend to always include few-shot examples in your prompts. Prompts without few-shot examples are likely to be less effective.” ↑
  6. 6IBM, What is zero-shot learning?, undated, read at source 12 Oct 2026“Zero-shot learning (ZSL) is a machine learning scenario in which an AI model is trained to recognize and categorize objects or concepts without having seen any examples of those categories or concepts beforehand.” ↑
  7. 7Palatucci, Pomerleau, Hinton & Mitchell, Zero-Shot Learning with Semantic Output Codes, NeurIPS 2009, read at source 12 Oct 2026“We consider the problem of zero-shot learning, where the goal is to learn a classifier f : X → Y that must predict novel values of Y that were omitted from the training set.” … “Some work by Larochelle et al. (2008) on zero-data learning has shown the ability to predict novel classes of digits that were omitted from a training set.” ↑
  8. 8Brown et al., Language Models are Few-Shot Learners, arXiv:2005.14165, 28 May 2020 (v4 22 Jul 2020), read at source 12 Oct 2026; the PDF gives the authors’ affiliation as OpenAI: “Zero-Shot (0S) is the same as one-shot except that no demonstrations are allowed, and the model is only given a natural language instruction describing the task.”↑
  9. 9Google, Prompt design strategies, read at source 12 Oct 2026“Few-shot prompts are often used to regulate the formatting, phrasing, scoping, or general patterning of model responses.” ↑
  10. 10Kojima et al., Large Language Models are Zero-Shot Reasoners, arXiv:2205.11916, 24 May 2022, read at source 12 Oct 2026“we show that LLMs are decent zero-shot reasoners by simply adding ‘Let’s think step by step’ before each answer.” ↑

IBM, What is zero-shot prompting? and What is zero-shot learning?, both undated, read at source 12 Oct 2026.

Palatucci, Pomerleau, Hinton & Mitchell, Zero-Shot Learning with Semantic Output Codes, NeurIPS 2009; Brown et al., Language Models are Few-Shot Learners, 2020; Kojima et al., Large Language Models are Zero-Shot Reasoners, 2022; all read at source 12 Oct 2026.

The comparison on this page (sorting letters for a new helper), the “In a sentence” line and the email example are this site’s own, not claims from these sources.

Free printable AI cheat sheets, and AI news every week.

Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.

SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK
aipromptgeneer://$ about this siteAI PROMPT GENEERAI explained in plain English. Every claim links to its source, with the date it was checked.NEWSLETTERWhat changed in AI each week, in plain English.SUBSCRIBE FREE ↗AI PROMPT GENEER · MACHINE ASSISTANCE DISCLOSEDEVERY ARTICLE CARRIES ITS OWN CHECKED DATE