How might Game Artificial Intelligence (AI) inform and influence learning experience design in the future?
What learning design used to be… and, sometimes, still is
To begin to answer that far-reaching question, let us consider what learning design or instructional design was, and in many cases still is. Before the days of learner agency, when didactic dinosaurs roamed the school corridors, lesson plans would map what had to be learned, how it would be taught, what resources would be required, and what time would be allowed. And those were the good ones; half of them didn’t even have a lesson plan. Mathematics had to be acquired in three-quarters of an hour fitted in between Geography, Gym, and lunch. A bell marked the intervals between chunks of teaching. Note: I don’t say between chunks of learning, because, in my case at least, learning didn’t happen.
Formative assessment was applied in the form of exercises and repetition, and summative assessment was made at the end with an examination. Punishment and ridicule were associated with failure as motivators. Prize giving in front of the parents distributed pride and shame in equal measure.
The game designer’s mindset — a growth mindset
To a game designer, this system seems a little odd. In games, success is rewarded by levelling up to face ever bigger challenges, and failure immediately presents other opportunities to succeed. If at first you don’t succeed explore the landscape for ways to try and try again. Players are not shown how to do that. Instead, they are presented with challenges, and the fun is in finding out for themselves how to overcome all the obstacles. Games tell you what needs to be done, they don’t tell you how. This approach ensures a high level of engagement. Challenges are often presented in the form of quests, and that in itself — like Jason and his Argonauts — suggests a big journey. If you reach Game Over, this is not the end, it’s just an opportunity to eat vegetarian pizza, drink sugar-free coke, and start over.
This is growth mindset (Carol Dweck article in Education Week), not yet-ness, and there really is no end, only opportunities for the future.
The future of learning design
To see into the future of learning design, one must see learning as a landscape. Distributed with a certain granularity in that landscape are nodes. At each node, there is a puzzle. Solving the puzzle rewards the student-player with an artefact, tool, or power that will be useful in the next part of their quest. These things they collect have value, and a collection of them is a student-player’s wealth. The quest — within the scope of the landscape — is of their choosing, but is stated at the setting forth, at the start: “Together we will travel to Mordor and destroy the One Ring”. The student-players know the objective, but they do not know how to get there, nor the struggles they will have to face along the way. They must find their way and they must find ways to defeat the bad guys, traverse the deserts, and span the canyons.
I am speaking metaphorically. I am not suggesting that the future of all e-learning is reduced to a video game. I am saying that under the hood of e-learning there will be an e-learning engine very similar to a game engine. This engine will be designed and built using Game AI techniques.
The relevance of game artificial intelligence and the future of education
Game AI is subject to various interpretations: Bartle, when he uses the term AI, means mobiles or non-playing-characters, fairies in the forest that help or taunt the traveller; Buckland describes Game AI as “the illusion of intelligence”; in its simplest form Game AI involves accessing a large database of options using if-else statements. In Vehicles, Experiments in Synthetic Psychology, Braitenberg shows how apparently complex animal and human behaviours arise out of the interaction of very simple internal structures.
Game AI also consists of several simple internal structures: databases; conditional statements; pathways, and configurable parameters. It is the putting together and the interaction between these structures that generates the underpinning intelligence commonly known as the game engine. Game engines can then support any number of modifications such as storylines, scenes, characters, graphics, music, and effects provided they remain in the genre and within the scope of the game engine. This has clear relevance to games-in-education as future teachers pour curriculum into the mixer.
Tools for the learning experience designer of the future
The learning experience designer of the future needs a tool very different to the old lesson plan. One such tool is graphing theory.
Buckman (Programming Game AI by Example) writes: “When developing the AI for games, one of the most common uses for graphs is to represent a network of paths an agent can use to navigate around its environment.” Each node has value, and each node is connected by one or more edges that have a cost. The value of the nodes along a narrative arc must eventually exceed the cost of the edges, or motivation will be lost. As the agentic student-player leaves a node, that fact is recorded in a learning record store (LRS). They solved a puzzle here, so the system knows they learned something. They proceeded along one of the edges to the next node, and they picked up embedded artefacts along the way.
Those embedded artefacts are what the didactic dinosaurs of the past used to call content. Games designers probably think more in terms of clues, tools, and powers for how to solve the next puzzle or overcome the next micro-challenge. The system records every incremental step along the way, forever. The system knows for example, that the student-player has acquired a certain tool and used it to solve a certain type of puzzle. Correctly analysed, interpreted, and presented, this data tells us a lot about the development and current state of the student-player. The system and its associated LRS is both omniscient and pervasive.
The immutable laws of nature don’t change. But the language we use to describe them, and the way in which we approach them do. The lesson plan of the past, the template, is a simplistic thing seen against a game engine. But the complexity is handled by a computer, and once built, the game engine is just a given thing. Now, the subject matter experts and the learning designers feed challenges, puzzles, and artefacts into the engine and crank her up. Employers no longer admire a candidate’s certificates, they go and look at the dashboard of the LRS and they see what strategies for effective gameplay that person has developed, and contemplate how they might apply them to new contexts of design, manufacture, distribution, retail, service, policy making, or administration.
The times are a changing — fast
The challenge for learning designers, in fact, for all educators, is to keep up with the pace of change. Noting the copyright dates of the suggested reading list (below) points up the true nature of this challenge, that time is ticking. Games-in-education conversations tend to be focused on educational value, or the lack of it, and fears around excessive screen time and isolation. The conversations learning designers need to be having are probably around theoretical underpinnings, and how they can be put into practice to create virtual world engines, incremental accreditation, and student-player profiles.
Suggested reading:
Bartle, R. (2004) Designing Virtual Worlds
Braitenberg, V. (1984) Vehicles: Experiments in Synthetic Psychology
Buckland, M. (2005) Programming Game AI By Example (Wordware Game Developers Library)
Stephen Lowe
Latest posts by Stephen Lowe (see all)
- Towards excellent user support - July 10, 2018
- Making a start with student data analysis - May 24, 2018
- LMS or LRS – It takes two to tango - April 10, 2018
