About ten years ago a friend of mine attracted quite a bit of negative publicity – including a harshly critical opinion piece in the New York Times – when his company developed an automated math tutor that, in a controlled study, gave better feedback and improved undergraduate students’ math skills more than human teaching assistants.

At the time, the tone of the press coverage was “Look at these horrible tech companies trying to put already-underpaid teaching assistants and adjunct professors out of a job!” However, today Harvard is using AI to teach computer science classes, Bill Gates is declaring that AI tutors can give every child access to a “really great” teacher, and 72% of HR leaders say AI is going to take over a good portion of their workforce training.

But while it’s true that AI can (and likely will) take over some work from human training instructors, coaches, and professors, that doesn’t mean AI must replace all human involvement in the learning process. In fact, when leveraged creatively, AI can complement and enhance human-led training workshops, courses, or coaching sessions. 

To give an example, my own company has been using AI-generated simulation exercises and role plays for workforce training, such as training charge nurses to manage hospital wards or giving tips to salespeople on how to position their products. But we don’t market those exercises as replacements for traditional workshops and coaching programs, but rather as tools that can be leveraged in these programs (the same way traditional e-learning and instructor-led training have long been combined in “blended” learning programs).

Specifically, we’ve found these types of interactive AI simulations and role plays can be leveraged in at least 3 different ways, with varying degrees of human interaction and involvement.

As any competitive athlete knows, there’s no substitute for doing lots of repetitions during training – Japan’s increasing dominance of international baseball is widely credited as being a result of the emphasis that the country’s baseball programs place on having players perform a massive number of repetitive practice drills (e.g.1000 bat swings!). Thanks to AI simulations, it’s now possible for people to get large numbers of reps in when practicing job skills from sales negotiations to patient triage to disaster response – which in the past could only be practiced in live training workshops.

And in addition to saving the logistical hassle of attending a live session, being able to simulate conversations with an AI is a huge benefit for beginners who want to practice their skills but get nervous or lack the resources to practice with others. Simply consider how many people that want to learn a language never do because they feel too embarrassed to try and practice it, or the number of people who don’t volunteer for role play exercises during customer service training.

What’s more, the amount of practice that you can get out of an AI is leagues beyond what a human facilitator can provide: no human coach, no matter how devoted, is going to want to do 10, 20, or 50 practice sales calls with a single trainee and will probably get irritated if the trainee occasionally says “Never mind – start over” or “Wait a minute – I got to take a call, be right back.”

And that’s just the limits that humans have to deal with during training – preparing simulations is equally time consuming. The simulations that we’ve trained Parrotbox to create in mere seconds would take a human hours… if not longer.

But that’s not the case with AI. The simulators we’ve developed can create a new simulation in seconds, which means that people can practice to their heart’s content.

The availability of AI-generated training simulations doesn’t actually mean there’s no value in people practicing skills with other humans. While some companies will undoubtedly try to use AI to eliminate human-led workshops as a cost-cutting measure, AI-based simulations actually work best in concert with human interaction, allowing individuals to practice “digital hands-on-training” at their own pace, before applying their skills in the field or with other participants in the training program.

For example, if a group of trainees is assigned to do conflict resolution simulations as homework, they can come together during their training session and discuss the scenarios they encountered. Exploring the differences in their experiences and discussing how they overcame them can ensure that everyone comes away with an appreciation for just how many different situations and experiences they could encounter on the job.

And, using solutions like Parrotbox.AI, training participants can save and forward transcripts of their sessions to other people – such as their coaches – and even allow for practice sessions to be observed in real time. This way, even if AI is facilitating the training simulation, human peers and instructors can still give personalized feedback on how a trainee performed during a simulation.

Finally, AI is great for group work during in-person sessions. Professors, coaches, and training facilitators have always used case-based learning in classrooms – taking trainees through various “what if” scenarios that involve group discussions, or “fishbowl” activities where one or two people do a role play while the rest of the group observes or gives feedback. And, with AI, facilitators don’t have to prepare these scenarios and participants don’t have to worry about playing the “non-player characters” (e.g., a patient in a hospital sim), everyone can simply focus on getting through the sim.

During live sessions, training cohorts can work on AI simulations as a team, with one person “driving” the simulation while the entire group discusses which decision to make next. This sort of approach is great for two things:

First, with multiple people brainstorming answers to a common scenario, it can open participants’ minds to multiple perspectives and a wider range of reasonable solutions. And discussing their decision-making process with others can also help learners recognize opportunities for improvement: for instance, two public safety officers comparing how each of them would respond to a particular incident then reconcile or learn from each others’ viewpoints.

Second, because scenarios can be generated instantly, it allows for “breakout” groups to walk through different scenarios – something few facilitators would have time to adequately prepare.

We just reviewed three ways that AI can help to improve how humans learn and train – and our team has been discovering even more applications as we work with clients introducing AI to their training programs. If your organization could potentially benefit from these types of interactive AI training resources, we would love to hear from you and collaborate.