What is a large language model?
The model type behind ChatGPT, Claude and Gemini — and the specific things it is unreliable at.
Knowledge hub
What is a large language model?
A large language model is a neural network trained on very large volumes of text to predict what text should come next. That single capability, at sufficient scale, produces fluent writing, summarisation, translation, reasoning and code generation across many tasks.
What it is actually doing
Predicting likely continuations. It has no database of facts to consult and no awareness of whether a statement is true — it produces text that is statistically plausible given everything it has seen.
That explains both the impressive fluency and the characteristic failure: a confidently written statement that happens to be wrong. Fluency and accuracy are separate properties, and the model optimises for the first.
What it is unreliable at
- Specific facts about your business, which it has never seen — fix with retrieval.
- Recent events after its training cutoff — fix with search or retrieval.
- Precise arithmetic and counting — fix by giving it a calculator tool.
- Knowing what it does not know — fix with grounding and an explicit refusal path.
What this means commercially
Almost every useful business deployment is a retrieval problem before it is a model problem. Get the right context in front of the model, require citation, and let it refuse. The model choice matters far less than most vendors suggest.
Talk it through before you commit
A discovery call is a working session on your constraint, not a sales pitch.
Related
Where this connects
Quick inquiry
Tell us what you're trying to build
A short note is enough. You'll hear back from the team, not a bot — usually within one working day.
- +91 83695 17339
- Palava City, Dombivli, Thane, MH 421204