Intro (creativity gateway)

     Rather than being content with a label of being limited only to entertainment, the beneficial task of making gaming a medium for endless productivity can have its success defined in diverse ways. It is perhaps most effectively illustrated by its ability to act as an interface & organizational framework for generative AI prompting (including fine-tuning & complex workflows), resulting in prompt parameters that are influenced by countless unique situations, producing dynamic context-based experiences shaped by a multitude of preferences, establishing an immersive relationship with games that positions them as meaningful productivity tools.

     The main parts of this article are about highlighting their usefulness in various categories: motivational, fitness, educational, and of course, AI integration. NPCs (non-player characters) can act as freethinking apps that serve different purposes. Different workflows can be carried out & behaviors varied when the appropriate characters or stimuli are brought together and collaborate to complete large, complex tasks. They can contribute with their own personality, creating increasingly unique products or results. Just like AI can challenge players in a game, it can translate original experiences into other data formats, such as quizzes.

     When it comes to something like basic knowledge, if you play a game about history like Age of Empires, you should leave that game with overlapping knowledge that would be gained by taking a history or social studies class. It can still be fun, but playing the game should be worthy of homework. In Age of Empires, there is background story for the campaign that isn't required reading, and the campaign quests are based on historic events.

Simulating Real-World Experiences

     Maybe you've learned lots of prompt modifiers or techniques and want a way to organize or store them, and to effectively recall them when appropriate. There are many ways to approach this that games have already explored & solved. There's no inventive prerequisite or imperative for trial and error with inefficient methods, or ones that work for some people but not others. Something as simple as a recognizable item on a shelf or a cooking recipe with templates for different tasks can be arranged conveniently, as though they were objects in a room, and not lines of text floating around someone's mind.

     Game worlds can provide a familiar, engaging, and addictive medium to interact with sophisticated algorithms and navigate complex data structures. Desirable outcomes can be measured and progress tracked with a game mechanism. Such integrated features empower LLMs to be expressive by using established preferences to communicate. Many app projects aren't successful simply because their solutions are saturated and redundant, but at least with games many people will find them fun regardless of how much unique utility they offer.


One particular example of exponentially getting value out of concepts that already exist is uploading a book, such as anyfavoritebook.pdf, and requesting a game be made out of the book. Settings can include:

  •graphics: existing game reference
  •objective/theme: puzzle/mystery requiring knowledge of book
  •gameplay: first person shooter
  •length: 20 minutes to complete
  •difficulty: expert knowledge of book is necessary to proceed, keen general gaming skill
  •creativity: within scope of book or expand on referenced concepts

     You would select which character you would play as, and there could be a gradient-based selection of how "open world" the game is allowed to be or how strict it must stay within the confinement of the events that take place in the book. If it is an action game you may be able to choose how generous they are with the weapons, such as a "first person shooter" with unlimited ammunition, or needing to find a few bullets and aim carefully or otherwise resort to a melee weapon or stealth. This would be part of a chosen difficulty for the game, based on required book knowledge & understanding in addition to general gaming skill.

Pursuing Substance

     New features that AI has to offer gaming aren't limited to entertainment. Responsive, adaptive difficulty can also be used to challenge players in many different ways that aren't limited to text. If you are communicating with NPCs, they can have their own avatars with freedom to express and evolve their respective personalities. Avenues already explored in gaming can be revisited in uncharted ways. Any data provided can be manipulated to provide statistical information in a graph or chart form, and subject to calculated ratings, opinions, and suggestions. Multitasking while working with numerous agents is also made easy and fun.

     When not mass distributed and sold, generating a personalized experience may avoid copyright concerns. An individual would simply upload the required data and play an unscripted game. If there is an expectation of it being polished before the first time playing it, the AI could interview the player, listing 3 options for how a game concept can be implemented, with the person basically going through a multiple choice checklist when it is not obvious what was intended. One way the AI comes in handy is the ability for the gameplay to measure progress and adapt the game to not be too easy or hard—manually coding branching paths would be a lot of unnecessary work and code.

Intuitive concept segues:

  •A hand grip tool can require squeezing it hard to get through doors in a game
  •Upload quotes and associate them with a particular NPCs persona
  •Answers to prompts can be implemented via relevant game mechanics
  •Characters can remind or make suggestions regarding aspirational goals
  •NPCs can come up with thought-provoking riddles & experience-based quizzes
  •Uploading .pdf or other files can be integrated into a character or setting

     Anything you can prompt can find its way into the game world in innovative ways. For example, if a given NPC agent is asked how would it perform a task, it would already have the prompt parameters associated with the game world, recent experiences with other characters, and any qualities unique to that character.

Summary

     Taking advantage of emerging technologies & features can be most efficiently achieved by exploring examples from gaming of different graphical user interfaces, integrated preferences, and parameter manipulation. Harnessing existing features can be better at this than simply adding LLM functionality, as organization and implementation can become complex. Referencing existing gaming concepts may also be the most personalizable path to beneficial collaboration with AI.

     As of this writing, some of these topics may be forward-looking for what LLMs will be capable of soon. But progress will only make this evergreen article more relevant. It's not only about what AI tools are capable of, but how they are used most effectively and enjoyably. An appropriate question to ask as more LLM & open-source features become ubiquitous, has been: what is your favorite way to use AI? Why would immersive fun be any less productive?

---