How mine sites are building training programs faster with AI
A new mine site cannot simply inherit the training program from another operation. Different regulations, equipment, procedures, and job requirements can all change what workers need to know before they step onto the site. The workforce itself may also bring a different combination of experience and skills.
That means training teams often have to create a program specifically for the operation in front of them. Traditionally, developing that material involved extensive research, scripting, video production, voice recording, editing, and review. By the time everything was completed, the project itself may have already changed.
That challenge is at the center of a Mining Now episode featuring host Jared Downey, Dan MacDonald of BIS Safety Software, Vincent Galati of BHP Jansen, and Mike Schwarz of MyZone AI. The discussion looks at how AI is changing course development, what remains unchanged, and where human approval and accountability still matter.
Who Is on the Episode
Dan MacDonald
CEO, BIS Safety Software
Vincent Galati
Lead Training Specialist, BHP Jansen
Mike Schwarz
CEO, MyZone AI
KEY TAKEAWAYS
- A new mine site needs training based on its own regulations, equipment, work scope, and workforce rather than simply reusing another site’s material.
- Faster course development can make adaptive learning, microlearning, and multilingual training more achievable within normal training resources.
- AI can support the development process, but people remain responsible for decisions, review, and final approval.
Onboarding Is a Major Challenge at a New Mine Site
BHP Jansen provides a clear example of the scale involved in preparing a new mining operation.
The potash mine under construction in Saskatchewan includes a production shaft and a service shaft located roughly a mile underground, along with a railyard, tailings management areas, and a port. Thousands of workers will move through the project during construction, with that workforce eventually carrying the operation through commissioning and into production.
Every worker needs the appropriate training before accessing the site. Requirements can vary based on the person’s role, responsibilities, and activities, while credentials also need to connect with other systems used for permits, room bookings, shuttle scheduling, and other business applications.
That puts the training team in an important position. Training records help establish who has completed the required training and who is qualified to perform particular work.
The process also begins much earlier than many people might expect. Contract requirements, project scope, schedules, and applicable regulations all contribute to the initial training needs analysis before construction is fully underway.
A New Approach to Building Courses
Traditional course development has been a lengthy process. Training teams would research the subject, write a script, produce video, arrange voice talent, edit the material, and send the finished course through review. Depending on the subject, that could take months.
AI is changing parts of that workflow.
Rather than beginning every project with new research, organizations can use documents they already have as the starting point. Existing policies, procedures, manuals, and other approved materials can provide the source information needed to develop a course.
The process can then move through several stages:
- Existing procedures, policies, and manuals are used as source material.
- An initial course outline is generated for the training team to review.
- Subject matter experts correct and refine the outline before development continues.
- Questions, images, narration, and other course elements can then be created around the approved structure.
- Additional languages can be incorporated into the development process instead of becoming an entirely separate project.
- People remain involved in reviewing and approving the material at each stage.
Vincent describes AI as another tool available to the training team when it addresses a specific problem. It does not take responsibility for the decisions that determine what workers need to know.
“It doesn’t augment our key decision making and it doesn’t augment accountability either. Those are two key factors.”
Vincent Galati, Lead Training Specialist, BHP Jansen
Capturing Knowledge Before It Leaves the Workforce
Some of the knowledge that contributes to a safe operation never makes it into a written procedure.
Experienced workers build knowledge over years of working with equipment, handling different situations, and understanding the practical realities of a job. When those employees retire, organizations can lose valuable information that was never formally documented.
The panel discusses the importance of capturing that knowledge before experienced workers leave. Subject matter experts can be brought into the course-development process and asked to share what they have learned over their careers.
This does not necessarily require lengthy interviews. A few structured conversations each month with employees approaching retirement can uncover useful information that might otherwise disappear.
That knowledge can then be reviewed and turned into microlearning or other training content that remains available to the workforce after the experienced employee has moved on.
Jansen will mobilize thousands of workers through construction and into operations, each needing site access training tied to their role.
Where the Human Still Has to Sit in the Loop
As AI makes training content faster to produce, the importance of reviewing that content becomes even greater. In mining, a course that is mostly accurate can still be a problem if it contains incorrect instructions.
Mike’s perspective is that software itself is becoming easier to reproduce, while years of industry-specific knowledge and the human review needed to keep information accurate remain much harder to replicate.
Shadow AI Is Already Inside Most Organizations
Employees are already experimenting with AI tools, whether or not their organization has formally approved a platform. The panel is candid about the risks this can create and the questions organizations need to ask before using these tools: where does information go, what can the tool access, does it use submitted information for training, and have the privacy settings been properly reviewed before the technology is put into use?
