Prototyping with AI
Completed listening, observing, communicating, discussing, and speaking at the Learning Technologies ’26 conference in London. There was one topic that dominated the Expo and almost the entire session:
Human Intelligence Vs. Artificial intelligence
Who’s winning? Is this a competition? What is hype and what is reality today? Where does it study? Are we making a difference? What is changing? What should change? Are we behind? Are people interested in measuring impact or measuring the illusion of impact? Can we still communicate as humans in the age of Artificial Intelligence (AI)? Here is my one-word definition of the experience: dialogue.
I wrote two screenplays. One of them was bad. But in between, for years, I was working on my job, creating good conversations.
Dialogue is a conversation between two or more people, or a written exchange between characters in books, plays, and films. It serves as a tool for characterisation, characterisation, and plot development, and can refer to a larger, collaborative exchange of ideas aimed at mutual understanding.
So, imagine, for a second, that we are actors in a movie. We all have a backstory, a belief system, a history of failures and successes, biases (known or unknown), etc. Some characters have human intelligence in our story, while others are artificial. We have a limited view of the world, past, present, or future. Dialogue takes place in scenes to further the plot. Every scene is important in the movie. As they move the plot forward, they reveal personality and help the characters grow.
Thinking about the author’s sunglasses
What Is Dialogue Not?
Speeches, downloads, mansplaining, speeches, content, information dumping, dashboards, Sharepoint sites…
Scene 1: Dinner for international speakers
Before the conference, some of the speakers and conference chairs met for an informal dinner. What did we eat? I don’t remember eating. But I miss the characters and the conversation we had. The discussion assumes the same goal of mutual understanding! Consensus does not mean complete agreement. You can completely disagree with someone but still have a conversation with them. But this only happens when there is at least some level of trust, respect, and openness. Conversation includes listening. Active and open listening. Not waiting for your turn to speak. Awaiting response.
We touched on mental safety, play, food, travel, and, of course, other topics related to learning. There were no slides, no tools, no next clicks. Building communication through conversations will remain important in the age of AI.
Consider two scenarios:
- Your boss sends you a well-crafted note about your project accomplishments. Short, concise, emotional, and perfect grammar. Besides, it is clearly written by AI.
- Your manager sends a note about the same accomplishment. It’s not perfect, but it took time and effort between two important meetings. May contain typos.
Many people would say that they prefer authentic personal messages and commissions. But us? There are AI promoters with product authenticity driving online traffic, chatbots rated as more empathetic than human doctors, or AI service agents replacing long wait times due to “unusually high call volume.”
I don’t have the answer, but I suspect that when collaboration is work, it works, and you don’t care about long-term relationships, AI will dominate the conversation.
Scene 2: Truth vs. The Hype
The current AI landscape feels like the land of Oz. On the hand, a magical illusion dominates LinkedIn: experts in every corner are multifaceted. Every decent learning technology vendor now offers AI-driven features, from content creation to simulation. While L&D is still working on agile engineering, some leaders have moved on to content engineering, and the rest of the world is building a chief of staff with OpenClaw.
Where is the result?
DX looked at AI and engineering results in a long-term study:
Many leaders feel that their organizations are lagging behind in the race to unlock the speed of AI-driven engineering. Retailer marketing and social media set expectations for 3x or 10x improvement. When leaders see modest results, they think something is wrong.
To provide that picture, DX analyzed the speed of engineering from November 2024 to February 2026 in a sample of 400+ companies where the adoption of AI increased significantly. We found a 10-15% increase in PR which is a real benefit, but less than most leaders expect.
The paper then goes into detail as to why the expected benefits of working with AI have so far not been met [1].
What about L&D?
There is a lot of research now focusing on the impact of AI on L&D. Research findings from RedThread Research, Egle Vinauskaite, Markus Bernhardt, and others, provide guidance on what’s happening in L&D (and beyond), and how to manage the future.
Speaking of management: My session was more about rapid modeling with AI tools. L&D always had a problem with rapid, iterative design to demonstrate working models. It used to require technical expertise and often IT support. Today, AI can speed up the process and enable learning professionals to test, learn, and learn quickly with prototypes. I defined this as a journey where you need a destination (a business problem or opportunity), a car (an AI tool that matches your need for cost, speed, and control), and a map of how to get there (not a static map in the old sense, like GPS directions on how to start a journey).
But if we let AI drive the process, and we just passively participate, it will be an expensive journey to learn how quickly we can go to places we never intended to be.
The truth is that AI is not a technology that L&D should “use.” At least, that’s not the only angle. And it’s hardly the beginning. It’s tempting to demonstrate the efficiency benefits of using AI to automate content creation, for example. My challenge to all L&D leaders is to move away from rapid content creation and efficiency measurement. And that doesn’t start with AI. It starts with understanding how we work today, and how we should work tomorrow:
- How are things done today? What is workflow?
- Who makes what decisions?
- Who is responsible for what outcome?
- How do you define the quality of an output? How do you look at quality?
- What is the expected outcome?
I know asking questions can feel like it’s slowing you down, but it will help speed you along the journey while minimizing the edges you might run into.
Episode 3: Why You Should Prototype, What Should You Prototype?
A common mistake is to treat a prototype as a cheap version of the real thing. These prototypes are often stuck in the prototype phase because they don’t measure up and don’t answer any questions (besides “can we build it?”).
The prototype is for learning. Learning something quickly and repeatedly. The prototype should focus on the most important part of the simulated information. If it’s your first AI chatbot to help employees, you don’t need to build a full app to learn that what it produces isn’t working for your audience. Play testing with real business problems and real users is important.
What if you learn that your idea doesn’t work? Well, save resources and time to build something that will do. I have seen “acquisition problems” for many applications in the corporate world because the team did not match the basic information. “If it builds, they will come” is not a strategy.
What Can You Prototype?
First, start with a business problem or opportunity worth solving. Efficiency is a simple ingredient, but it can backfire. At one point, I created an automation that takes text and creates a PowerPoint deck from content in minutes. I thought I saved hundreds of HeH (human equivalent hours). Kind of. It helped us continue to build a dynamic voice presentation quickly. Also, make sure there is a business case for the future, not just for the current phase.
Second, start with the end in mind: who is your audience and how will they reach a solution. The prototype doesn’t have to be perfect, but for scalability, you need to keep your final deliverable in mind while making a prototype version of it.
Who is the target audience?
- You
It can be a practical application that helps with professionalism or quality assessment. For example, if you are responsible for evaluating the quality of a test question, it is a good target for a skilled AI agent. If you haven’t built an AI agent yet, but want to improve the User Experience in the eLearning courses you create, that may also be a realistic goal. - Your peers
What if you could solve problems in your team’s current workflow? What if you could create something that adds to that process or replaces some of the things? For example, if you use xAPI, you can create a statement builder for your team that follows your standards and generates login code. If you’re still playing with SCORM, you can build something similar. - Your organization
What if you can solve different workflow problems? What if a useful tool could help others make their work easier, faster, or find relevant information faster? What if you could ditch the old, outdated training courses and replace them with an effective real-time support assistant? - Employees (“students”)
What if you could embed dialogue within the learning experience? Or a simulation tailored to the role, location, and previous skill level? Sometimes, you just need to “innovate” in the sense of logic: you already have an LMS that authenticates users and stores data (with SCORM cmi statements), to use a useful tool that is compatible, customized, and effective, by implementing a deep link. Of course, a dedicated web server with single sign-on would be better, but for now, you can clone the tool.
I’m talking about access: I suggested in my session that, no matter how small the first prototype may be, everyone should start with programming. Specifically, planning the entire solution (not just the prototype) in a product requirement document (PRD). All LLMs know exactly what PRD is, and can build the foundations for it. You can then extend this document as one of the project artifacts.
No matter what AI tool you use (I alternate between Windsurf, Claude Code/Coworker, and Github Copilot), this basic PRD will help make decisions and set a tight scope for the prototype with the final solution in mind. All of the above is related to one thing: dialogue. Meaningful, iterative conversations between humans and AI.
Now, Go and Build Something!
PS If you’re wondering what the picture represents (beyond the reflection of the sunglasses), you’ll need to investigate the Banksy sculpture in the background. At first, it’s supposed to be about blind patriotism, with someone blinded by the flag going into free fall. To me, it brings uniformity to AI. Take charge, learn, and experiment. You can just follow the influencers.
Photo credits:
- The image within the body of the article is provided by the author.
References:
[1] AI and the speed of engineering: A longitudinal analysis
