A large language model (LLM) is the engine behind assistants such as ChatGPT, Claude, Grok and Gemini. It is trained on huge amounts of text and writes by predicting the next word, which makes it fluent but able to state wrong facts with confidence.
AItopic
2 minread
3related terms
In one line
The engine behind ChatGPT, Claude, Grok and Gemini. It is trained on huge amounts of text and writes by predicting the next word, which is why it is fluent but can state wrong facts with confidence.
An LLM does not look facts up the way a search engine does. It has learned patterns of language, so when you ask it a question it produces the most likely sounding answer. Often that answer is right. Sometimes it is a confident guess, which is called a hallucination. This is why the same question can get slightly different answers each time you ask it, and why wording your prompt clearly changes the result.
Modern assistants add tools around the model so it can search the web, read your files or connect to your ATS. Those tools give it real data to work from, which cuts down on made up answers, but you still need to check anything that matters. LLMs are measured in tokens, which are chunks of words. The amount of text a model can hold in view at once is its context window, and the cost of most AI tools is tied to how many tokens you use.
Why it matters to recruiters
Most new recruiting software runs on an LLM under the hood, from AI outreach writers to interview summaries and job description generators. Knowing how the model works tells you where it is strong, such as drafting, summarizing and rewriting, and where it needs a human check, such as facts about a candidate, salary figures and anything legal.
An example from a recruiting desk
With a job description and five resumes pasted into an assistant, a sourcer asks for a short summary of each candidate against the must haves. The summaries take a minute instead of an hour. One summary says a candidate holds a security clearance, which is not on the resume. The sourcer checks, corrects it, and from then on asks the model to quote the resume line behind each claim.
Tips that help
Give the model the source text and ask it to answer only from that text.
Ask it to quote the line it relied on so you can check claims quickly.
Keep candidate personal data out of tools your company has not approved.
Save prompts that work so the whole team gets consistent results.
Mistakes to avoid
Pasting a model’s answer into a client email without reading it.
Asking an LLM for salary data or legal rules and trusting the reply without a source.
Frequently asked questions
Is ChatGPT a large language model?
ChatGPT is an assistant app built on top of OpenAI’s large language models. The model is the engine; the app adds the chat window, memory, file uploads and tools such as web search.
Why do LLMs make things up?
They are built to produce likely text, not to check facts. When the answer is not in what they learned or in what you gave them, they can fill the gap with something that sounds right. Grounding them in your own documents reduces this.
Can I use an LLM to screen resumes?
You can use one to summarize resumes against clear criteria, but a person should make the decision. Several US laws now cover automated tools in hiring, so check your state and local rules before letting software rank or reject applicants.
