Let the game keep growing.

Seeing learning through game creation

This session drew our attention to six areas. These questions help us revisit specific moments and guide further observation. They are not scores of learners’ abilities, nor evidence that those abilities have already improved.

Clear communication: how can others understand the game you imagine?

Can teammates and AI understand the same request? Can learners explain the player’s actions, goals and outcomes? When a generated result differs from the idea, first check whether the request was clear, then consider whether the implementation went wrong.

Collaboration: how do different ideas become a shared decision?

Who suggests ideas, who operates the tools, and who tests the game? How are disagreements resolved? Notice whether everyone participates in decisions, as well as whether they work together on the game.

Careful design: do the rules hold up when someone plays?

Do starting, playing, winning, losing and restarting fit together? What happens after repeated clicks, when resources run out, or when a player reaches a boundary? Use playtesting to reveal gaps, then add a rule that can be tested.

AI’s limits: generating a game does not guarantee it works as intended

Which requirements did AI implement, which did it miss, and which features only appear to work until someone tests them? Distinguish unclear requests, undecided rules, platform limitations and generation errors. Avoid blaming learners for every problem or leaving the judgement to AI.

Interest: what do learners enjoy playing, and what do they want to make?

Which kinds of gameplay do learners bring up themselves? Which details are they willing to keep improving? Record what they enjoy experiencing and what they want to keep creating separately, leaving room for interests to change.

Attention: when are learners engaged, and when does their focus shift?

Observe playing, discussing, waiting for generation, troubleshooting and repeated refinement separately. Record the task and situation. One distracted moment does not establish a lack of focus, and a long time looking at a screen does not necessarily mean productive engagement.

How the session unfolded

1. Play an example gameWe first played the teacher’s AHA Space game, exploring and interacting with a small world from a player’s perspective.
2. Discuss favourite kinds of gamesWe talked about the games we enjoy and the experiences we would like to create.
3. Form teamsLearners formed groups around shared interests and began imagining a game together.
4. Explore game-making platformsWe discussed different platforms, then tried an online AI coding platform.
5. Make a game with AI coding toolsDescribe an idea to AI, generate a version, preview it, and keep adjusting.
6. Playtest and reflectPlaying and discussing the games showed that they needed more work after the session.

The session provided a starting point for continued creation. The following activities are suggestions for later work, not a claim that every group completed them that day.

Start with one action a player can try

Explain what the player does, how they know whether they have succeeded or failed, and how to try again. Leave extra maps, items and complex features for a later version.

Your first brief for AI

We want to make a small browser game in the [game genre] genre. The player uses [one main action] to achieve [goal], and fails when [failure condition]. Explain the controls at the start, show the result at the end, and allow a restart. Support both computers and phones. Leave out [features to defer] in the first version. First, restate the rules and ask us about anything unclear.

Change one thing at a time

During playtesting, [what actually happened] occurred at [location or action]. We want [expected result]. Please change only this and keep the rest of the gameplay. Explain how we can check the change afterwards.

AI’s output does not decide what the game ought to be. The group chooses the rules, checks the results and decides which changes to keep.

Rotate three roles within the group

One person describes the rules, one uses the platform, and one playtests and records problems. Swap roles in the next round so that everyone helps make decisions and check results.

Next time, improve one thing

1. Ask another player to tryGive them a task and observe before explaining. Note where they stop, misunderstand something or want to try again.
2. Choose the problem that most affects the experienceFor example, a player cannot find how to start, a phone button is hard to tap, or the game cannot restart after a loss. Fix that first, then consider adding features.
3. Ask the same player to try againGive them the same task and compare what happens before and after the change. If it has not helped, investigate again.
4. Share an updateProvide a working game link, controls, a one-sentence description of the gameplay, what you changed and what remains unresolved. Do not include personal information or account credentials.

Let a creation open a door into the small world

AHA Space explores the theme of rational inquiry and a gentle approach to life. We plan to connect learners’ games to it: players explore the space, complete tasks and discover hidden entrances to other games.

An entrance invites real players to try a game; it is not a trophy reserved for finished work. Games can keep evolving, with more entrances added as new work is shared. We hope these connections encourage continued refinement. Whether they sustain engagement is something to observe over time.

The planned entrances

• Fish in the pond: collect a fishing rod from the little girl’s room, then explore the pond beside the pavilion.
• Survival at sea: unlock tasks involving a campfire and a small boat.
• Operation Gunnerside: the Vemork raid — complete the mechanical sentry challenge to obtain mission clues.

Before linking a game, we check its link, basic controls and suitability of content. Works in progress are identified honestly; a first submission is not treated as a finished game.

Game entrances become available as services launch and are updated. Describing an entrance here does not mean every service is already available.

From spaces to games: code and exercises

Instructions are primarily in Chinese. Reconstruction targets Windows; a full run on a fresh machine has not been verified.

Further reading

Read this public article for more observations and reflections on the session.

Return with these three questions

Which part do players most want to try again? Where do they still need the creator to explain? What would your group like to change first in the next version?

Alongside a more complete game, we hope to see a clear explanation of why a change was made.