Dive Training Leaders Need AI Guardrails That Protect Safety and Judgment
By: Dr. Gleb Tsipursky
Dive training leaders face a different AI adoption challenge than most professionals. A plausible but wrong output can affect course planning, student communication, equipment guidance, incident documentation, and decisions made around real people in demanding environments. The useful question is therefore not whether an instructor, dive center, or training team should use AI. The question is where AI can save time without weakening the human judgment, standards, and accountability that safe diving requires.
Start With Low-Risk Administrative Work
The safest early uses sit away from real-time safety decisions. A dive center might use AI to draft a course reminder, turn a long policy into a student checklist, create several versions of a social post, organize frequently asked questions, or summarize feedback after a class. An instructor might use it to outline a lesson, suggest examples for explaining buoyancy concepts, or rewrite technical language for a beginner audience.
These tasks still need review, but mistakes are easier to catch before they cause harm. They also produce visible value. Staff can save time on repetitive writing while keeping final control over what students, customers, and other professionals receive.
Separate Assistance From Authority
AI should support preparation and communication rather than act as an authority on fitness to dive, emergency response, decompression decisions, equipment condition, environmental hazards, or student readiness. Those judgments depend on context, current standards, direct observation, and professional responsibility.
A useful internal rule is simple: AI may propose, organize, compare, or summarize. A qualified person must decide, verify, and approve. That distinction protects instructors from gradually treating a confident response as a substitute for expertise.
Build Verification Into Each Workflow
A general instruction to “check the AI” is too vague. Training organizations need a verification step tied to each use case.
For course communications, compare dates, prerequisites, locations, required equipment, and certification details against the official course record. For lesson materials, compare claims against current NAUI educational material and approved standards. For equipment-related content, verify information against manufacturer documentation and established training guidance. For incident summaries, compare the draft against original notes and remove assumptions the record does not support.
The reviewer should know exactly what to inspect and which source controls when the AI output conflicts with it.
Protect Student and Business Information
Instructors and dive operators handle personal details, medical information, waivers, payment data, incident reports, and internal business records. Staff should never place sensitive information into an unapproved public AI tool simply because the tool makes summarization convenient.
Organizations need a short written policy that identifies approved tools, prohibited data, required anonymization, and the person to contact when uncertainty arises. Clear boundaries make responsible experimentation easier because staff know where they have permission to proceed.
Use AI to Strengthen Instruction, Not Flatten It
Good instruction responds to the student in front of the instructor. A generic AI-generated explanation may sound polished while missing fear, confusion, physical limitation, language needs, or an unsafe misconception.
AI can help instructors generate alternative explanations, scenarios, knowledge checks, and visual analogies. The instructor should then choose and adapt the material based on the learner, the course, and observed performance. The value comes from expanding the instructor’s options, not standardizing every student interaction.
Treat Experienced Skeptics as Safety Resources
Some of the most cautious instructors will notice failure modes that enthusiastic adopters miss. They may question whether an output reflects current standards, whether a checklist omits an exception, or whether a simplified explanation creates false confidence.
Invite those instructors to test proposed workflows. Ask them to identify where an error could affect safety, trust, or compliance. Their scrutiny can improve the guardrails and make the eventual process more credible across the organization.
Measure Outcomes Beyond Time Saved
Time savings matter, but they are not enough. A dive organization should also track correction rates, student confusion, staff confidence, consistency of communications, privacy incidents, and whether instructors continue using the workflow after the initial experiment.
A useful pilot can start with one repetitive task for two or three weeks. Compare the old and new process, document common errors, revise the prompt and review checklist, and decide whether the workflow deserves broader use. This turns AI adoption into a practical improvement process rather than a technology demonstration.
Create a Short AI Use Protocol
A workable protocol can fit on one page:
- Define the permitted task
- Identify the approved tool and prohibited data
- Name the authoritative source for verification
- Assign a qualified human reviewer
- Record recurring errors and update the workflow
- Escalate any safety, standards, medical, legal, or incident-related uncertainty
The National Institute of Standards and Technology describes its AI Risk Management Framework as voluntary guidance for managing AI risks and incorporating trustworthiness into how organizations design, use, and evaluate AI systems. That broad approach fits dive training well because it treats governance, measurement, and ongoing review as part of adoption rather than as paperwork added later.
Keep the Human Role Visible
The strongest AI workflow makes human responsibility clearer. Students should know that instructors remain accountable for teaching, evaluation, and safety decisions. Staff should know that reporting an AI error helps improve the process rather than creating blame. Leaders should show that efficiency never overrides professional standards.
Dive training organizations can gain real value from AI in communications, preparation, administration, and learning support. They will gain that value sustainably when they start with bounded tasks, protect sensitive information, verify outputs against authoritative material, and preserve the instructor’s role as the final decision-maker.
Adapted from: The Psychology of AI Adoption at Work: From Resistance to Results (Georgetown University Press, 2026)
Dr. Gleb Tsipursky, a behavioral scientist called the “Office Whisperer” by The New York Times, helps tech-forward leaders stop overpaying for AI while boosting engagement and innovation. He serves as the CEO of the AI consultancy Disaster Avoidance Experts, and wrote eight books, including The Psychology of AI Adoption at Work: From Resistance to Results (Georgetown University Press, 2026).
