Developer English guide
English for Working with AI: Prompts, LLMs and Pair Programming in English
September 22, 2026
Why the best AI prompts are in English, the vocabulary you need, and how to talk about AI work with your team.
Key Takeaways
- English prompts produce better results in most LLMs because models are primarily trained on English text.
- Knowing the right AI vocabulary lets you discuss your work clearly with your team.
- Pair programming with AI requires specific English phrases to get useful, precise output.
- Always verify AI output: hallucinations are common, especially in technical details.
- AI documentation is in English. Learning to read it faster is a high-leverage skill.
Why prompts work better in English
The majority of text used to train large language models is in English. This means that for technical tasks, such as writing code, reviewing architecture, or debugging, English prompts tend to produce more precise and detailed responses.
This isn't a rule without exceptions. Models have improved significantly in Spanish and other languages. But for technical content specifically, the gap still exists. A prompt like “Write a Go function that implements a rate limiter using the token bucket algorithm” will consistently outperform a Spanish equivalent in most current models.
Practical test you can run right now:
Ask the same technical question in English and in Spanish. Compare the depth and accuracy of the response. For most complex coding questions, the English response will be more detailed and technically accurate.
AI vocabulary for developers in English
These are the terms you need to discuss AI work in English with your team. Many have no direct Spanish equivalent and are used as-is in Spanish-language tech conversations.
| Term | Español | Example sentence |
|---|---|---|
| Prompt | Instrucción / consulta | Write a clear, specific prompt to get a useful response. |
| Token | Token | GPT-4 has a context window of 128k tokens. |
| Context window | Ventana de contexto | The context window limits how much text the model can process at once. |
| Hallucination | Alucinación | The model hallucinated an API method that doesn't exist. |
| Fine-tuning | Ajuste fino | We fine-tuned the model on our internal codebase. |
| Embedding | Embedding / representación vectorial | Embeddings let us search semantically across our documentation. |
| RAG | RAG (Retrieval-Augmented Generation) | We built a RAG pipeline so the model can answer questions about our codebase. |
| Inference | Inferencia | Inference cost is the main operational expense for LLM applications. |
| Temperature | Temperatura | Set temperature to 0 for deterministic outputs. |
| System prompt | Prompt de sistema | The system prompt defines the model's behavior and constraints. |
| Few-shot prompting | Prompting con ejemplos | Provide 2-3 examples in your prompt to guide the model's output format. |
| Chain of thought | Cadena de razonamiento | Ask the model to think step by step for complex reasoning tasks. |
| Grounding | Fundamentación | Grounding prevents hallucinations by anchoring responses to real data. |
| Guardrails | Barreras / restricciones | We added guardrails to prevent the model from generating harmful content. |
| Multimodal | Multimodal | GPT-4V is multimodal: it can process both text and images. |
For the full AI vocabulary guide, see AI Vocabulary for Developers.
How to pair program with AI in English: useful phrases
Effective AI pair programming requires specific English phrases. The more precise your prompt, the more useful the response. These phrases work across Copilot, Claude, and ChatGPT.
“Write a function that...”
Starting point for any code generation task.
“Refactor this to use...”
Ask for a specific pattern or data structure.
“Explain what this code does.”
Get a plain-English explanation of unfamiliar code.
“What are the edge cases here?”
Ask the model to think about what your code might miss.
“Is there a more idiomatic way to write this?”
Get language-specific best practices.
“Why did you choose this approach?”
Understand the reasoning, not just the code.
“This looks like it has an O(n²) complexity. Can you optimize it?”
Challenge the model to improve time complexity.
Example: a full pair programming session in English
You: “Write a Python function that validates an email address.”
[AI generates code]
You: “What are the edge cases this doesn't handle?”
[AI lists edge cases]
You: “Refactor this to use a compiled regex for better performance.”
[AI refactors]
You: “Is there a more idiomatic way to write this in Python?”
How to discuss AI work with your team in English
When you discuss AI-generated work with your team in English, you need phrases that communicate both what the AI did and what you verified. Transparency about AI involvement is becoming standard practice.
“I used Copilot to scaffold this but reviewed every line.”
“The model hallucinated this part. I need to verify it.”
“Let's add a human review step before shipping this AI output.”
“The prompt needs more context about our domain.”
“This is a false positive from the classifier.”
Notice the pattern: these phrases communicate your role in verifying and adapting AI output. They show professional judgment, not blind reliance on the tool.
AI tools documentation is in English: how to read it faster
AI tools like the OpenAI API, Anthropic Claude, Google Gemini, and LangChain all publish their documentation in English. Reading it efficiently is a skill in itself.
Strategy: examples first, explanation second
Skip to the code examples first. Once you see what the code does, the surrounding explanation makes much more sense. This works for any technical documentation in English.
Common doc patterns you need to recognise:
Deprecated
This feature still works but will be removed in a future version. Stop using it now.
Experimental
The API may change without notice. Don't use this in production.
Breaking change
Upgrading to this version will break existing code. Read the migration guide.
Under the hood
An informal phrase meaning 'how it works internally'. Usually precedes an explanation of implementation details.
For more technical vocabulary used in developer documentation, see Coding Acronyms Every Developer Should Know.
Related articles
Ready to practice your English at work?
Lingua-e has interactive exercises built around real developer conversations: standups, code reviews, retrospectives, and more. Practice until it comes naturally.
Try Lingua-e for free
Written by
Roxana LafuenteLingua-e's founder
Roxana Lafuente is a software engineer with 8+ years of experience. At the beginning of her career, even though she had already passed the First Certificate in English, she still froze every time she had to speak up in the daily standup. That was a gap nobody was fixing. After 2,000+ standups, she figured out what actually builds fluency: practice that looks like your real work. She built Lingua-e so other developers wouldn't have to take the long road to feel confident working in an international development environment.