Artificial intelligence is becoming an increasingly useful everyday tool, but how far should fire safety professionals rely on it when interpreting standards, codes of practice and technical requirements?

Q: Can I use AI to fact-check the application of standards and codes of practice?

The Short Answer: No. Due to the fundamental way artificial intelligence operates, it cannot be relied upon as an authority for factual information, legal compliance, or code interpretation.

The Long Answer

Assessing life-safety risks requires professional judgment, deep subject-matter expertise, and verified sources. While AI is a helpful assistant for summarising long documents or drafting emails, it is unreliable as a technical reference tool. In fire safety work, the correct answer often depends on the building, the use of the premises, the people who may be at risk, the fire precautions already in place, and the exact wording of the relevant law, standard, guidance or code of practice. These matters require professional judgement and proper checking against current source documents.

AI falls short in supporting fire safety professionals for a few core reasons:

• It Predicts Language, Not Truth: AI models don't "know" facts. They use massive datasets to recognize patterns in words and statistically calculate the most likely next word in a sentence. It prioritizes what sounds convincing over what is objectively true.
• "Hallucinations" Look Real: AI models are notorious for completely inventing clauses, facts, quotes, and legal sources. Because the output is written in a confident, professional tone, these errors are incredibly difficult to spot without manual verification.
• Outdated or Missing Data: Most AI models have data cutoff dates. Furthermore, they may not have access to proprietary or paywalled documents—meaning crucial British Standards (BS), codes of practice, or precise legislative updates might be missing from their training data entirely.

One of the main risks is that an answer can sound confident even when it is wrong. It may refer to outdated requirements, a misunderstanding of the context, or an answer that does not apply to the building or situation being considered. In some cases, it may also refer to information that has not been properly checked against the current standard, code or legal requirement.

The fire safety landscape is highly complex, spanning distinct purpose groups and evolving legislation and guidance. As a general rule, if you aren't competent enough to fact-check the AI's output yourself, you shouldn't be using it for that task.

It is often helpful to talk through complex scenarios with trusted mentors or peers; an AI model is not an appropriate substitute. Identifying, researching, and verifying appropriate sources of information is a core competency of a Fire Risk Assessor.

Generally, if someone is not able to check the answer themselves against the proper source material, they should not rely on AI for that task. Knowing what to check, where to check it, and how to apply it correctly is a core part of professional competence in fire safety.