Did You Know Why LLMs Struggle with Numbers?

Did you know that tools like ChatGPT and other large language models (LLMs) don’t actually understand numbers? To a model trained on language, the symbol “10” is just another token, no different than the word “anchor” or “storm.”

This means that while LLMs are excellent at generating text, they can sometimes make mistakes when asked to calculate, count, or convert data. They aren’t built as calculators; they’re designed to predict the next word in a sequence. That’s why users sometimes encounter surprising or inaccurate results when relying on AI for math-heavy tasks.

Why This Matters in Maritime

In the maritime industry, precision is non-negotiable. A single incorrect figure can cascade into major compliance, financial, or operational issues. Imagine:

• A small error in emissions reporting can lead to penalties or non-compliance findings.
• Miscalculations in fuel cost comparisons are impacting operational budgets.
• Inaccurate data in audit reports raises red flags during inspections.

In such high-stakes environments, relying on LLMs to “guess” numbers introduces unnecessary risk.

Smarter Ways to Use AI in Data-Heavy Maritime Workflows

So does this mean AI isn’t useful for numbers? Not at all. It simply means we need to use it strategically.

Here are three smarter ways to use AI in maritime contexts:

Use AI to generate code that does the math

Instead of asking the model to calculate directly, have it produce a Python script that runs the calculation reliably. This way, AI supports the task while accuracy is guaranteed by the code execution.

Connect AI to trusted tools and APIs

Link the model to live data sources such as weather systems, fuel pricing databases, or regulatory repositories. This ensures that any numerical information is sourced and validated, not guessed.

Leverage AI for pattern recognition, not precision

LLMs excel at identifying trends across large sets of unstructured data. For example, they can:

• Spot recurring deficiencies in inspection reports.
• Highlight common causes of incidents across fleets.
• Summarize large datasets into actionable insights.

By combining these approaches, maritime teams can get the best of both worlds: smarter workflows powered by AI, and reliable accuracy anchored in trusted systems.

NAVIREGO’s Approach

At NAVIREGO, we understand both the capabilities and limitations of generative AI. That’s why our platform doesn’t ask AI to “do the math” blindly. Instead, we focus on embedding AI into maritime workflows where it adds the most value:

• Inspection data analysis: AI identifies patterns, trends, and recurring issues, enabling QHSE teams to address systemic risks.
• Audit readiness: AI surfaces relevant regulations quickly, ensuring audit data is accurate and compliant.
• Operational efficiency: AI assists in drafting and structuring reports, while calculations are validated through trusted sources.

This balance ensures that maritime professionals benefit from AI-driven speed and insights, without sacrificing the accuracy their work demands.

Final Thought

LLMs like ChatGPT aren’t calculators. They’re powerful assistants for language, context, and pattern recognition, but they need guardrails when it comes to numbers. In maritime operations, where precision is paramount, that distinction is critical.

By combining AI’s strengths with domain expertise, validated data, and secure connections, NAVIREGO helps maritime organizations unlock the benefits of AI while keeping full confidence in their numbers.

Ready to explore smarter ways of using AI in maritime?
[Talk to NAVIREGO today] (https://www.navirego.com/contact) and discover how we turn AI’s potential into reliable, compliance-ready solutions.