When it comes to top executives and sales leaders who bring in new business, most organizations spend lavishly on professional development. Microsoft, IBM, Goldman Sachs, and Unilever hire coaches who might charge as much as a thousand dollars an hour to help executives manage stress and improve their decision-making, and leadership skills. Google holds its annual sales retreats at luxurious resorts in places like Hawaii or the Mediterranean, with gourmet meals, famous keynote speakers, concerts and parties. And elite medical institutions like Mayo Clinic spare no expense on physician training, with their state-of-the-art simulation center that includes realistic patient mannequins, virtual reality theaters, and replica operating theaters and emergency rooms, all of which cost millions of dollars to maintain.

However, when it comes to the other 95% of their workforce, organizations tend to be stingy, relying on less effective but lower-cost measures like e-learning or one-off webinars conducted via Zoom. The average online professional education course for a nurse, consisting of some videos and quizzes, costs $20 to $200 per hour, while a typical safety course for a construction worker might cost $50. And if you asked most corporate training departments about the gross disparity in investment between executive training and frontline worker training, they would likely cry poor about limited training budgets… or just laugh in your face.

However, AI-based learning tools have the power to close this “training gap” and provide entry-level employees and middle managers something closer to the high-touch training that the people atop the organizational pyramid receive. Instead of being expected to watch a video then get to work, AI-generated training allows workers at every level to practice walking through realistic scenarios to practice their skills and receive guidance when they need it on the job.

And if this sounds like a far-off, techno pipe dream – it’s not. While it might not match a Hawaiian sales retreat or the Mayo Clinic’s elaborate practice hospital, there are already affordable solutions to give workers the benefit of simulation-based training, personalized coaching, case-based assessments, and robust on-the-job support using AI.

Let’s look at some specific examples of how organizations can achieve this, today:

The Power of Simulation and Role Play

Traditionally, staging realistic simulation-based training exercises was almost as time and resource-intensive as carrying out job tasks for real. Consider the case of a large city organizing a simulation exercise for mass casualty events (e.g. earthquakes, terrorist attacks): these exercises might involve hundreds if not a thousand plus hospital staff, firefighters, law enforcement officers, and other workers, plus actors to play disaster victims. Or think about Harvard Business School, where students spend an entire semester pretending they are already high-powered corporate executives, walking through “what would you do in Company X’s situation?” examples based on exhaustively researched, real-life case studies that often take years to research and prepare. On a smaller scale, organizations might bring their leaders together in a hotel conference room to play Dungeons & Dragons style games about supply chain management or crisis management, complete with elaborate probability tables and dice rolls to determine the success of their actions, with an outside consultant acting as referee.

If these measures sound expensive and time consuming: they are. However, the expense is often justified by the results. According to the National Training Laboratories, learners retain 75% of knowledge gained through simulation and role play, compared to just 10% from reading and 5% from lectures. The Journal of Applied Psychology and SHRM report that simulation training improves skill acquisition by 20-35% over other methods. And the International Journal of Education Research found that learners who engage in role play perform 40% better on practical assessments.

Of course, there are reasons simulation-based training isn’t widely used. Bringing together a dozen busy executives for a workshop, let alone several hundred emergency responders for a disaster rehearsal, is a massive commitment, and developing complex video-game-like simulations for self-guided training can take so long that the simulation is obsolete before it reaches its intended audience.

For example, when our company did a financial planning exercise for healthcare CFOs in the mid 2010s, it took weeks to pull together mock financial data for various types of healthcare systems. Likewise, it took a month of collaborating with subject matter experts to develop the background info on a fictional flooding incident for an emergency management communications workshop.

However, with AI, the preparation time required to design scenarios can be reduced to milliseconds, and the time required for computer-based simulation development can be effectively eliminated. When provided with sufficient, high-quality reference materials and clear parameters for how to structure and facilitate a role play activity, an AI can generate all the technical details, characters, and situations required in less time than it takes a live facilitator to open a PowerPoint presentation. This means that every employee can now engage in on-demand simulation practice, receiving immediate feedback from the AI (without the need for an outside consultant or manager to play “Dungeon Master”) and honing their skills in a highly realistic and immersive, yet risk-free environment.

And once the parameters of a simulation are dialed in, the AI can spawn infinite variations on a theme (e.g. a doctor communicating with a patient or a customer service rep handling a complaint), allowing learners to get far more variation and repetition than they could with human-facilitated exercises.

To see how this looks in practice, check out the following demos of healthcare and financial services training that our company developed with Parrotbox.AI, a tool specifically designed to deliver AI generated simulations.

The Benefits of Personal Coaching

There’s a reason why top executives spend tens or hundreds of thousands of dollars for 1-to-1 coaching.

According to the International Coach Federation (ICF), 70% of workers who receive coaching improve work performance, and 61% improve managerial skills. The Human Capital Institute reports higher employee engagement and 20% lower turnover among coached employees. Bersin found that coached employees are 33% more likely to fill critical leadership positions, and the Journal of Occupational Health Psychology notes reduced stress among coached employees. Additionally, the Corporate Executive Board found that coaching leads to 50% faster onboarding with improved performance.

However, qualified coaches are in short supply, making it difficult to scale coaching programs while ensuring consistency. In-house managers and experienced staff usually have limited time to coach others on top of their own job duties. And when it comes to hiring outside coaching talent, credible executive coaches can charge tens or hundreds of thousands of dollars per year, while even less specialized coaches can cost anywhere from $65 an hour for a nurse educator to $350 per hour for a sales trainer to $500 per hour for a technical expert in fields like data science or engineering. This makes it financially unfeasible for most organizations to provide personal coaching to all employees.

AI can help overcome these barriers, not by replacing human coaches, but rather by reducing the effort required for experienced professionals or consultants to support coachees on the job. For example, an ESG (Environmental, Sustainability and Governance) consulting firm can “ground” an AI agent with documentation of their recommended best practices and frameworks, and have it answer straightforward questions per the coaching methods endorsed by organizations like the ICF, such as offering prescriptive versus collaborative advice depending on the apparent knowledge level of the user. These agents can often act as an on-demand source of basic information and guidance, then “escalate” more complex issues to a human coach at the user’s request.

To give an applied example, here is a “virtual coach” designed to help smallholder farmers in low-income countries apply climate-smart farming practices, and another to help salespeople position their products for customers.

The Effectiveness of Case-Based and Free-Response Assessments

Let’s be honest: multiple-choice questions (MCQs) are a poor measure of true understanding. Most of us have memories of cramming for standardized tests then forgetting everything we memorized the day after, and there is an entire cottage industry around helping people learn strategies to game the system and identify correct answers by looking for patterns in how questions and response options are phrased, without necessarily grasping the actual material.

Free response, case-based assessments – where learners are presented with a situation and expected to describe their response at length with no predefined response options – are more effective than multiple-choice quizzes. Deloitte found that free-response assessments are 20% more accurate for gauging mastery of complex topics. The Journal of Educational Psychology reports that case-based learning and assessment boost performance by 30% on critical thinking tasks. Butler & Roediger (2008) found that free-response questions increase retention by 25%, and the Journal of Applied Testing Technology notes that case-based assessments are 50% better predictors of future performance.

Despite their effectiveness, free-response assessment methods have been underutilized due to the time and expertise required to design effective assessments and ensure consistent scoring. However, given AI’s impressive ability to process natural language, AI quizzing tools can do roughly as well as human graders – provided they’re given a clear “rubric” and examples of how to analyze and score responses. To see some examples, here are some scenario-based assessment generators that draw their material from Canadian anti-money laundering guidelines and US food safety regulations:

The Advantages of Point-of-Performance Support

In manufacturing, it’s long been established that – for most situations – being able to quickly produce and deliver products “just in time” in response to demand is better than manufacturing a large number of products “just in case” only to have them gather dust in warehouses.

The same principle can also apply to human knowledge: organizations might have staff spend a whole week in a workshop practicing tasks they might not need to perform right now on a “just in case” basis, only to have them forget everything before having an opportunity to apply it. A better approach is usually to find ways to deliver information on a “just in time” basis by offering point of performance support resources like job aids and knowledge bases.

And evidence shows that point-of-performance (PoP) support often outperforms traditional training approaches. IBM found that PoP tools increase productivity by 50% compared to traditional training alone. Employees access PoP resources six times more than traditional training resources, and PoP tools increase retention by 60% more than traditional training (ATD). Deloitte reports that PoP learning and knowledge management systems accelerate time to competency by 30%, and Bersin notes that PoP learning and KM systems increase engagement by 40%. Additionally, ASTD found that PoP reduces training costs by 30%.

However, the investment required to create and curate resources and integrate PoP tools into existing systems and workflows has been a barrier. And oftentimes, organizations compile excellent libraries of support materials only to have them sit unused in difficult-to-navigate intranet sites and network folders, just like unbought shoes or snack chips in a warehouse.

AI can help overcome these barriers by providing scalable, user-friendly PoP support, allowing employees immediate access to the information just by asking an AI copilot a question, then letting the machine do the searching on their behalf.

Overcoming Barriers to Best Practices with AI

Most learning professionals will openly acknowledge that simulation-based training, personal coaching, case-based assessment, and point-of-performance support beat out traditional e-learning and workshops, only to sigh, shrug their shoulders in defeat and mutter, “But who’s going to pay for that?”

However, today’s AI-based learning tools eliminate these excuses, making it possible for every worker – not just the C-suite executives and high-profile superstars – to get the training and support they need and deserve. Because, while much has been made about AI’s ability to perform certain tasks, perhaps its ultimate contribution is empowering people.