Where Are We Now

See, try, do. That’s the typical sequence of events in a training journey. Work through a module or course that outlines a topic. Complete an application exercise, maybe with a little feedback. Then, get out there and start doing it. We’re pretty good at the first part. L&D professionals have been honing that skill for a few decades at least. But more often than not, our application falls flat — or worse, get skipped entirely. That means our learners go from didactic, one-way modules (let’s be honest, that’s how the majority end up) straight into their daily jobs.

Application is hard to do well. And it’s even harder to do well at significant scale. Good application draws significantly from the broader concept of practice. Several decades of research on practice in areas like music, athletics and even video games help inform our understanding of what makes practice effective.

What Makes Good Practice

Good practice has three key elements: sustainment, challenge and feedback loops.

1. Sustainment: In isolation or as a one-off experience, practice loses much of its effectiveness. Giving something a rough run-through is not the same as deliberate practice. Focused sessions of practice, sustained over time, lead to the best results. This lines up well with a favorite concept in L&D, the Ebbinghaus forgetting curve. We know from research and from real-world experience that learning comes from repetition and re-application. Sustainment is what changes practice from an event into a process, leading to durable learning.
2. Challenge: Identifying the appropriate level of challenge is critical to deriving meaningful results from practice. If something is too easy, learners lose interest and end up in a “going through the motions” mentality. If something is too hard, learners can become frustrated and give up. Designing training that lives at the appropriate challenge point — where a learner is always being asked to perform at a level just beyond their current level of proficiency — is incredibly difficult to achieve at scale but represents one of the best ways to design for growth.
3. Feedback Loops: Practice on its own can generally lead to improvement. But feedback is a powerful multiplier that significantly enhances the improvements from practice. The most effective feedback is timely, relevant and constructive. Feedback reinforces performance and provides important contextual details that aid in memory retention. Without feedback to correct poor performance, practice can sometimes lead to learners developing habits around the wrong behaviors.

Knowing what makes for good practice and finding ways to incorporate it into your training programs are two different things. We create simulation-style scenarios that ask learners to make choices based on what they’ve learned. But after they’ve gone through it once, they know the right answers, removing any challenge. Alternately, we set up role-play exercises during instructor-led trainings (ILTs) where learners need to model behaviors based on what they know. But once the ILT is done, the opportunity for role play is over, removing any sustainment.

And both examples struggle with scale. Writing compelling simulations is a time-consuming exercise for a team with limited capacity to develop training. Providing personalized feedback to every participant in an ILT would take up too much valuable training time when there are other items on the agenda.

The Arrival of New Tools

The barriers of time, cost and scale aren’t new. What is new is our growing toolkit for addressing them.

Generative artificial intelligence (AI) doesn’t solve these challenges outright, but it introduces new ways to approach them. When used thoughtfully, it can help sustain practice over time, adapt challenge levels dynamically and provide timely, personalized feedback — three elements that have historically been difficult to deliver at scale — all while keeping the human in the loop.

Here’s how generative AI can address some of these obstacles to help create best-in-class practice and application.

• Sustainment: One of the biggest advantages of generative AI is the variability it offers. With a single prompt, you can create a role-play experience that mimics a real-world situation with a conversation that shifts and adapts to whatever the learner says. Without the confines of a pre-written script or dialogue tree, this single experience becomes a repeatable practice opportunity that can sustain learning over time.
• Challenge: It’s fairly simple within an AI prompt to give instruction as to the level of intended difficulty, but that doesn’t differ very much from traditionally built training experiences. The true value generative AI brings is in adaptive challenge. With the right prompting, an AI-powered training experience can modify the difficulty dynamically to provide the appropriate challenge level for individual learners. This wouldn’t make sense in every situation (for instance, a certification where every learner needs to see the same content), but for skill-based practice, this is an exciting way to challenge learners with material at their skill level.
• Feedback: Even the best coaches and facilitators have limited time to provide feedback. Taking some of that burden off the human and using generative AI to provide the first round (or rounds) of feedback can lead to the time spent with a coach becoming more valuable. If a learner has practiced before and received AI-generated feedback to cover the basics (based on well-designed rubrics and expectations), then when that learner gets time with their coach, that time can be spent on the nuanced types of feedback that humans are so good at (and where AI can still struggle).
• Psychological Safety: Generative AI-driven practice can serve as a more private and “safe to fail” approach. If you’ve ever role played during a training event, you know that there’s still pressure to perform well. Talking to an AI-powered persona can feel easier and like there’s less pressure to get it right. Even knowing that you can repeat an exercise as much as you need can help alleviate some of that concern.

Generative AI isn’t the end all, be all of practice design. But it’s the most promising way we’ve had to remove the excuses that keep practice out of training. The more chances we give people to try, fail and adjust, the more capable they, and our organizations, become.