Query
In artificial intelligence and database systems, a query is essentially a request for information or computation made to a system. While the concept originates in databases — asking for rows that satisfy specific conditions — in AI the notion has expanded.
- In information retrieval, a query can be a natural language search typed into an AI-powered engine (“What are the best restaurants nearby?”).
- In machine learning models, a query often refers to the input provided to the model to generate an output (e.g., prompting a language model or sending an image to a vision classifier).
- In vector databases and embeddings (e.g. image embeddings), queries are expressed as vectors, and similarity search retrieves the closest items.
The term also takes on a broader role in reinforcement learning (querying the environment for a reward signal) or in prompt engineering (where queries are carefully crafted to elicit reliable responses).
Thus, “query” in AI is not just a command — it is the bridge between human intent and machine reasoning, shaping how information is retrieved, interpreted, or generated. A query is more than a request—it’s the starting point of interaction between a user and an intelligent system. In AI, queries define what the system should care about and how it should allocate its computational resources. Without a clear query, even the most advanced model has no direction.
There are also distinctions between structured and unstructured queries. Structured queries (like SQL statements) are precise and machine-readable, while unstructured ones (like natural language prompts) require interpretation. Modern AI models thrive in the latter scenario, translating vague human input into actionable computation.
Queries also play a central role in evaluation. For example, in information retrieval benchmarks, performance is judged by how well a system responds to a set of queries. In the age of large language models, prompt engineering has become the art of designing queries that balance clarity, specificity, and creativity—shaping not only outputs but also user experience.
🔗 Reference:
- Manning et al. Introduction to Information Retrieval (Cambridge, 2008)
