Skip to content

Tag: AI

They earned my silence

Published by Paolo on 12/11/2024

Maybe it’s because I started on a new project. Maybe it’s because a company that canceled the position when I was the last candidate republished the position again. Maybe it’s because I am tired of influencers. Or maybe it’s because they use content to train shitty algorithms to produce empty content faster.

Maybe it’s because I don’t want to constantly check out likes and notifications there. Or maybe it’s just because I got bored.

I decided to commit to (at least) a period of silence on LinkedIn. I will continue to write here and on my Substack, though.

Creativity and patience

Published by Paolo on 14/06/2023

For me, there is a direct relationship between creativity and patience. Ideas need to rest before being properly evaluated. Teams need to have the space to make their own journey and thoughts to make a game happen.

Most games never get published. This is due to many factors. But, a good pre-production phase helps mitigate the risk of not seeing the light.

I’ve read many articles explaining how AI tools help speed up the pre-production stage of a game. Some say that companies can also create content faster. I am very skeptical on this point.

In the pre-production phase, a team measures its potential toward a concrete challenge. The AI tools promise to give us concept art of a pretty high standard in minutes. We can also create stories and document templates. We can get quick code snippets.

But then we’ll find ourselves having to edit here and there. This editing process is different from the process that created successful games.

Since when did we decide that faster is better?

A good dish takes time to cook. A good vertical slice or demo, too. People need time to make meaningful connections, the sparks that ignite the engines. If we entrust this process to machines, then we end up working for the machine.

I enter my prompt and await the results. I review and analyze them. I iterate with these results by introducing more prompts. I review everything and make my changes. Instead of me acting and creating, it’s like I’m making corrections to an assistant. And it’s one of the worst assistants because it doesn’t actually think!

Fail faster is good advice, but it doesn’t mean we have to rush things. If something not created by us fails, it will be more difficult to grow. We will have no memory or connections that will make us understand which steps need to improve.

When did we decide that jumbled datasets are better than looking for references?

People need the process of searching for references to achieve creative goals. While the result of a prompt may appear to have excellent quality, it is still a mindless mixing of elements.

Our urge to have “the thing” now causes us to end up feeding a machine that will create something average. It makes us disperse in a mass.

The process that created the hit games that are on everyone’s lips works differently. There are two types of goals, project goals and personal goals. Every maker must have time to reflect. This time is invested in looking for references and organizing them. The same goes for an artist, a writer and a programmer. If this process is skipped in the name of speed, we will be acting like monkeys. Can we make something good? Just by chance.

Is it possible to use these tools in a healthy way?

The quick answer is no because datasets are a sophisticated intellectual property assault.

For the extended answer, imagine that there is no ethical/legal problem. Assume that the datasets are completely legitimate.

These tools can be used to unlock meaningful internal conversations for the team.

If I, a game designer, have to communicate some concepts to artists, these tools can facilitate my work. If a producer is briefing game writers, these tools can help estimate the number of words to use.

AI tools can help us learn to communicate with people belonging to other departments.

There is a direct correlation between the time a team works together and their odds of success. We should foster this necessary time with patience.

  • Instead of thinking about speeding up critical passages, let’s improve cross-department communication.
  • Instead of trying to get to the end faster, let’s improve our understanding of how everything contributes to it.

AI to improve my workflow

Published by Paolo on 05/12/2022

Recently, I began a new collaboration that I hope will yield positive results. I apologize for not writing in a while due to the overwhelming amount of work I have been facing.

act like you are an interactive romance story

I am currently testing the ChatGPT tool and find it incredibly fascinating. I believe that this type of technology can help me become more productive and improve my abilities.

Using ChatGPT, you can write prompts and generate well-done, yet basic, content. To create truly human and original content, you will need to add your own input and edit the generated content.

Overall, I believe that using ChatGPT will save me hours of work and allow me to focus on other important tasks.

The World is ready for the TikTok of Puzzle Games

Published by Paolo on 25/05/2022

Few time ago I used to work on a match-3 game for a strong IP. The game never saw the light, but I was working for a company dedicated to design and production of many match-3 games for different IPs.

There was an interesting behavior we observed from Players: many of them jumped from game to game looking for brain training, fun and so on. I had a lot of match-3 games installed on my device at that time. When I wanted to really empathize with those Players, I finished all lives in one game before of jumping to another.

When you run a game app you have to wait 30-90 seconds before of seeing the very first screen. Then you probably receive a lot of special offers and pop-ups distracting you from your purpose: to play a match-3 session.

If I had data and people, I would create an infinite feed of puzzle levels where the Player can scroll and jump from game to game in a quick way, just like TikTok is offering video content.

The feed is a great metaphor

Today everyone has confidence with the feed metaphor. A feed is fast and offers always novelty. Also, intruding pop-ups can be substituted by feed elements.

  • Swipe, Swipe, Swipe and Play this
  • Swipe, Swipe, watch ad to get power-ups
  • Swipe, purchase the special pack

Mobiles are tecnologically advanced nowadays and can support multiple 2D levels on the same scene with optimization. Corporates like King, Peak, Jam City, Playrix and Zynga have a tremendous amount of Player’s data and can use that to train an artificial intelligence.

Artificial intelligence would help a lot with level design. From one side, AI can help with serve the right mechanics, challenge and visuals to the right audience. From the other, AI can help with the Player Generated Content. 

Imagine create your own levels and sharing them with the game's community. Immagine getting all lives and power-ups that the Players use on your levels. I think we are definitely ready for that!

But let’s go in order.

The art of liveops

When you have a live game you constantly work on four main fronts:

  1. FTUE: onboarding and tutorials for every feature of your game
  2. Player Progression: control and balance the growth of your Players inside of your game
  3. Game Quality: improve flows and optimize the cognitive load of your game
  4. New Features: design and deploy novelties to keep your Players engaged

Doing good liveops is working on those four fronts at the same time. A significative workload is taken by the content and level design for every update. Usually, you have people dedicated to create everything using the editors that are provided by the developers. Then every piece is tested intensively, repeating the level or the sequence over and over to double check everything is ready.

The legacy of Super Mario Maker

I love Nintendo. I have to say it. Maybe their business choices are different than we expect, but they have this magical aura to me that surrounds almost everything they deliver.

I think that if you are reading these words you probably know what Mario Maker is. Anyway, I leave you a video in case you don’t:

  • Create a level for any Super Mario game you know
  • Beat your own level 3 times
  • Share it with the Community (become Miyamoto)
  • Play levels made by other Players

Well played Nintendo! Now you don’t need to create a lot of levels to deliver a new Mario. People will do it, for fun!

Super Mario Maker was a success, people love to create levels for their favorite game. If I imagine to design a new game where Players will create their own levels, anyway, I would have some serious doubt that a new intellectual property can actually generate that impact.

My point, then, is simply that Nintendo managed to save hundreds of man hours in designing levels. When your game is good, people will engage in creating content.

Case Study: Modl.AI

Recently on Linkedin I saw an interesting post by Jamie Clarke:

Link to the original post

That reminded me when I was working on a match-3 game, managing a small team of level designers. The process was:

  • Review the Levels Beat Chart (4 hours)
  • Decide the mechanics to introduce (4 hours)
  • Discover all the possible skill atoms and discuss on a new set of levels (8 hours)
  • Design the levels (24 hours)
  • Test the new levels and leave comments (16 hours)
  • Iterate on the levels and repeat all the checks (16+ hours)
  • Play the levels over and over to estimate the fail rate and the number of movements for the level (16+ hours)

It was a great job and my lazy mind always thought: “can we PLEASE make SOMETHING more automatic?”

Well, maybe now we can with companies like Modl.AI.

Game Pitch: Match-3 TikTok!

Imagine deliver to the Players of a modern match-3 game a level editor using the help of the artificial intelligence. I really believe that is possible to do that, and that should be as easy as sharing a video on TikTok.

  • Select a type of challenge, mechanics and specific skill atom. For example: Create a 2 boards big level with Chocolate and Licorice.
  • AI Prepares the level for you and shows you the level automatically generated
  • A level editor helps you personalize your level, the tiles spawners and everything
  • Beat the level 3 times to prove that it’s not impossible to beat
  • Decorate your level: put a background, stickers and some social network stuff that people love to share!
  • Share it with the community
  • All the lives and power-ups invested by other Players on the levels you create will be your!

As final note, it would be better to have a successful game already. This can be a feature for top grossing match-3 games out there.