Companies around the world are pouring resources into AI training, but a persistent question remains: after the courses end and the certificates are issued, does anything actually change in how people work? Samsung’s large-scale effort to reskill employees offers a useful case study. The company has approached AI fluency as a top-down priority, not a one-off learning event. The underlying lesson for HR leaders is straightforward: training can open the door, but only leadership and follow-through can make employees walk through it.

Training Is Only the First Step

Most HR professionals have seen the pattern. A company announces an AI skills initiative, employees complete a few modules, and managers report that the workforce is now ‘AI-ready.’ Months later, however, spreadsheets are still managed the old way, meeting notes are still written by hand, and the new AI tools sit unused. The problem is not necessarily the quality of the training. It is that education alone rarely changes established work habits.

For training to translate into practice, employees need immediate opportunities to apply what they have learned. Short, scenario-based exercises during the course are not enough. They need real tasks where using AI is relevant, safe, and expected. This is where HR teams can design follow-up activities: team-level AI pilots, guided practice sessions, and project debriefs that ask people to reflect on what worked and what did not. Without these reinforcement loops, knowledge fades quickly and old routines reassert themselves.

Leadership Makes the Change Believable

Samsung’s reskilling push has been described as a top-down effort, and that choice matters. When executives and managers actively use AI in their own work, they send a message that the change is real. Employees are far more likely to experiment with unfamiliar tools when they see leaders doing the same thing, making mistakes, and adjusting. Christine Josephine of Samsung Indonesia has emphasized that leadership makes change believable and helps it stick. That observation captures a core dynamic in organizational behavior: people watch what leaders do more than they listen to what leaders say.

This does not mean managers must become AI experts. It means they need to model curiosity, ask questions informed by data or AI outputs, and visibly integrate AI into their routines. If a leader sponsors an AI training program but continues to demand reports in the same old format, employees receive conflicting signals. The safest interpretation is usually that the training is optional. When leaders change their own workflows first, they give employees permission to do the same.

Measuring Whether Anything Actually Changed

The question ‘Did anything actually change?’ should be answered with evidence, not enthusiasm. Completion rates and quiz scores are weak indicators of real transformation. HR teams need to define what successful adoption looks like before training begins. For example, do they expect customer service teams to resolve inquiries faster, recruiters to screen candidates more consistently, or marketing teams to produce first drafts with AI assistance? Once those expectations are clear, organizations can track relevant metrics: time saved on repetitive tasks, quality improvements, employee confidence, and actual usage of approved tools.

Measurement also creates accountability. If leaders review adoption data and discuss it with their teams, AI use becomes part of normal performance management rather than a short-lived campaign. Recognition for early adopters, updated templates that include AI steps, and regular check-ins all help turn new behaviors into standard practice. The goal is not to monitor employees like a surveillance tool but to understand whether the intended change is taking root.

For remote and hybrid teams, the challenge is even greater. Distributed workers do not have the same casual exposure to how colleagues and managers use AI. They may complete training in isolation and then return to a home office where no one is visibly modeling the new behavior. Remote-first organizations need to create explicit moments for AI experimentation, such as shared ‘show-and-tell’ sessions, chat channels for prompts and results, and leaders who narrate their own AI use in asynchronous updates. For flexible staffing platforms such as XMF, where talent may be spread across time zones and client environments, that principle is especially relevant: distributed teams require deliberate signals to make new skills stick.

Ultimately, Samsung’s experience is a reminder that AI transformation is not a training problem alone. It is a leadership and change-management challenge. Organizations that treat AI upskilling as a one-time program will likely see temporary awareness but little lasting change. Those that pair training with visible leadership, application opportunities, and meaningful measurement have a much better chance of answering yes when someone asks whether anything actually changed.

Originally published by XMF, inspired by publicly reported industry news.