AI, Job Cuts and the Future of Fire Investigation

Every week there seems to be another headline about AI, layoffs and the “end” of professional work as we know it. Atlassian recently confirmed structural job cuts as it reshapes itself for an “AI-first” future, saying it retained people with skills needed to thrive in that model. Across Australia and globally, other companies are making similar moves, and the public conversation is quickly shifting from “Will AI affect jobs?” to “Whose job is next?”

It is a fair question for fire investigators too. We work in a profession built on analysis, documentation, technical reasoning and expert opinion. On the surface, those are exactly the kinds of tasks AI appears ready to assist with. It can summarise interviews, sort photographs, draft timelines, search standards, compare data sets and even help structure reports. Used properly, that is powerful. But assistance is not replacement.

Fire investigation is not just information processing. It is the disciplined application of science, experience, scene interpretation and expert judgment to messy, imperfect, damaged and often incomplete evidence. The authorities in the field make that point clearly: fire investigation is a formal process of determining origin, cause and development using a science-based methodology, and the work must be carried out by competent, qualified people applying informed judgment. That distinction matters.

AI can organise data. It cannot walk through a fire scene and smell the difference between fuel load and contamination. It cannot meaningfully assess whether a witness is confidently wrong, evasive, traumatised or simply mistaken. It cannot independently recognise when the absence of evidence is a product of fire damage, suppression activity, scene disturbance or failed collection. And it certainly cannot own the opinion that ends up being tested in a courtroom.

That last point is becoming more important, not less. Courts are already responding to generative AI. In New South Wales, court practice notes now say generative AI must not be used to draft or prepare the content of an expert report without prior leave of the court, and if leave is granted, the expert must disclose what part was prepared using AI and what system was used. Similar Australian guidance also warns that special caution is needed if AI is used in documents representing a witness’s evidence or opinion.

That should be a wake-up call for every expert witness, including fire investigators.

The legal system is effectively drawing a line: AI may be a tool, but it cannot become the author of expert opinion. That makes sense. Expert evidence is not valuable because it is grammatically polished or fast to produce. It is valuable because it is the independent opinion of a person with specialised knowledge who can explain, defend and stand behind their reasoning.

And that is precisely why AI cannot replace fire investigators.

Our profession is built on more than output. It is built on accountability.

A fire investigator must be able to explain why one hypothesis was accepted and others were rejected. They must show how they moved from scene observations to analysis to conclusion. The scientific method is not optional window dressing in modern fire investigation; it is central to defensible origin-and-cause work and to expert testimony. The literature is clear that investigators ignoring or departing from accepted methodology face greater scrutiny, and courts have repeatedly examined whether fire experts actually followed a reliable scientific process rather than leaning only on experience or unsupported assumption.    

AI does not solve that problem. In some cases, it may make it worse. If investigators become lazy with AI, the risk is not just bad writing. The risk is false confidence. A polished paragraph generated in seconds can easily create the illusion of certainty where none exists. It can smooth over evidentiary gaps, overstate conclusions, or subtly import assumptions that were never actually proven. In a field where one weak conclusion can affect insurers, manufacturers, property owners, criminal defendants and grieving families, that is dangerous.

So where does that leave us?

In our view, AI will absolutely change fire investigation. It should. There is no virtue in wasting expert time on tasks a machine can genuinely assist with. AI can help investigators work faster, search broader, identify inconsistencies, manage larger document sets and reduce administrative drag. It may improve consistency in report structure, help cross-check citations, assist with scene inventories and support early case triage. For firms handling large volumes of work, those gains are real.

But the value will come from using AI to elevate human expertise, not substitute for it.

The future fire investigator will likely need a broader skill set than the investigator of the past. Technical fire knowledge will still be essential, but so will digital literacy, data discipline and a clear understanding of where AI can help and where it must stop. Investigators will need to know how to use AI critically, how to verify every output, how to protect confidentiality, and how to make sure their opinions remain their own.

In other words, the future belongs to investigators who are both scientifically grounded and technologically literate.

That is a very different message from the one often pushed in the layoff headlines. The real lesson from the Atlassian cuts and similar moves elsewhere is not that humans are obsolete. It is that routine, repeatable and easily systemised work is under pressure. Fire investigation has routine components, yes. But the profession itself is not routine. It is interpretive. It is evidentiary. It is context-heavy. It is ethically demanding. And when it reaches litigation, it is deeply personal and highly scrutinised.

If anything, AI may increase the importance of genuine experts. As more low-quality analysis, templated opinions and AI-assisted shortcuts enter the broader professional ecosystem, courts and clients are likely to place even greater value on investigators who can demonstrate real methodology, real independence and real judgment.

The future of AI for fire investigators is not a future without fire investigators. It is a future that will reward the ones who know the science, understand the limits of the technology, and are willing to own every opinion they sign.

Because in the end, AI does not give evidence. We do.