AI in Aviation Safety Management: Part 5 – Practical Use Cases

Written By Currie Russell

Where AI Can Help Today — And Where It Shouldn’t (Yet)

Over the past several weeks, we’ve explored the foundations of AI in aviation safety management—what it is, how it should be governed, how data shapes its effectiveness, and the human factors risks that come with it.

By this point, most safety leaders are no longer asking “What is AI?”
They’re asking something far more practical:

“What can we actually use this for—today?”

The answer is both encouraging and cautionary.

You don’t need a fully integrated system.
You don’t need to upload your entire SMS database.
And you don’t need to overhaul how your organization operates.

But you do need to understand where AI can genuinely help—and where it has no place (yet).

Start Here: AI as an Assistant, Not a Decision-Maker

Before getting into specific use cases, it’s important to anchor one principle:

AI should support your SMS—not run it.

AI is effective when it:

  • Accelerates routine work

  • Improves consistency

  • Prompts better thinking

AI is risky when it:

  • Replaces judgment

  • Makes decisions

  • Operates without oversight

If your Safety Management System is the structure, AI is simply a tool within it—not the system itself.

Where AI Can Help Today (Low Risk, High Value)

Most airports already have the core elements of an SMS in place:

  • Reporting systems (paper, spreadsheets, or platforms like Vortex or Wombat)

  • Investigation processes

  • Hazard registers

  • Corrective action tracking

The challenge is rarely the absence of a system.

The challenge is capacity, consistency, and insight.

That’s where AI can begin to add real value.

1. SMS Guidance and On-Demand Support

Even experienced SMS managers don’t have every regulation, procedure, or best practice at their fingertips.

AI can act as an on-demand reference tool:

  • Structuring investigations

  • Explaining risk assessment approaches

  • Providing examples of hazard categories

  • Offering reminders of what “good” looks like

Used this way, AI helps reduce reliance on memory and supports less experienced team members.

But it’s important to be clear:

AI can guide—but it cannot validate compliance.

That’s still the role of your organization—and where experienced partners like Acclivix continue to add value.

2. Investigation Support (Not Investigation Replacement)

Investigations are one of the most resource-intensive parts of an SMS—and one of the most inconsistent.

AI can help:

  • Draft interview questions

  • Organize investigator notes

  • Suggest potential contributing factors

  • Turn rough notes into structured reports

It can also challenge thinking:

“What factors might I be overlooking?”

This is where AI shines—as a thinking partner.

But it must not:

  • Determine root cause

  • Validate evidence

  • Replace investigator judgment

A poorly conducted investigation, even if well-written, is still a poor investigation.

3. Hazard Identification and Risk Thinking

Hazard registers often become static over time.
Teams focus on what has already happened, rather than what could happen.

AI can help expand that thinking:

  • Brainstorm hazards in specific operational contexts

  • Explore “what-if” scenarios

  • Identify precursors to known risks

For example:

  • Winter operations

  • Low visibility conditions

  • Airside vehicle movements

This supports a more proactive SMS, rather than a purely reactive one.

4. Building Better Corrective Action Plans

Corrective actions are frequently one of the weakest parts of an SMS:

  • Too generic

  • Not tied to root causes

  • Difficult to measure for effectiveness

AI can assist by:

  • Structuring actions (immediate, short-term, long-term)

  • Suggesting ways to monitor effectiveness

  • Linking actions back to underlying issues

For example:

“Based on this root cause, what would strong corrective actions look like?”

The result is not a final answer—but a better starting point.

5. Documentation and Communication

SMS outputs take time:

  • Investigation reports

  • Safety bulletins

  • Internal briefings

  • Toolbox talks

AI can significantly reduce that burden by:

  • Drafting content

  • Simplifying technical language

  • Translating materials (particularly valuable in Canadian operations)

This allows safety leaders to spend less time writing—and more time engaging with their teams.

The Big Question: Do We Have to Share Our Data?

For many organizations, this is where interest in AI stops.

And understandably so.

Safety data is sensitive.
Operational data is sensitive.
And trust—internally and externally—is critical.

The good news:

You do not need to upload your entire SMS database to benefit from AI.

In fact, you shouldn’t.

Practical approaches include:

  • Using de-identified summaries

  • Describing scenarios rather than sharing raw reports

  • Removing names, identifiers, and sensitive details

For example, instead of uploading a full occurrence report:

“An aircraft was pushed back and struck ground equipment due to miscommunication between ground crew and flight crew…”

This allows you to gain insight without exposing your data.

The key is not integration—it’s intentional use.

Where AI Should Not Be Used (Yet)

Just because AI can assist in many areas does not mean it belongs everywhere.

There are clear boundaries.

AI should not be used for:

  • Final risk acceptance decisions

  • Regulatory compliance sign-off

  • Sole-source investigations

  • Real-time operational decision-making

Why?

Because these require:

  • Accountability

  • Context

  • Experience

  • Judgment

AI has none of these.

Where Acclivix Fits

AI can help you do more with your SMS.

Acclivix helps ensure you’re doing the right things, the right way.

That includes:

  • Designing effective, auditable SMS processes

  • Strengthening investigation and risk methodologies

  • Validating outputs against regulatory expectations

  • Helping organizations integrate tools—like AI—responsibly

AI can enhance your system.

But it cannot build, validate, or sustain it.

A Practical Way Forward

You don’t need a formal AI strategy to begin.

Start small:

  • Use AI to support one part of your SMS

  • Keep a human in the loop

  • Be intentional about how and when it’s used

Because the real question isn’t:

“Should we be using AI?”

It’s:

“Are we using it deliberately—or not at all?”

And in a system built on continuous improvement, doing nothing may be the bigger risk.

Currie Russell
Next

IA dans la gestion de la sécurité aéronautique : Partie 5 – Cas d’utilisation pratiques