Is making games for me? 5 questions before you spend a dime

Is making games for me?

It can be, and knowing how to code is not what decides it. Today AI writes the game's code from what you describe. What decides it are five questions about your reason: taste, patience, time, size, and the nerve to show your game. They are our reading, and each one comes with a fact from someone who has been through it.

Picture the scene. You have always wanted to make a game. You are standing at the door, before starting, with the usual doubt: "but I do not know how to code". That is exactly the question that was left off the list.

The five questions measure your reason, not your talent. Answer them honestly, preferably in writing, before you subscribe to anything or pay any store. None of them is an exam; they help you see where the work will weigh.

Why is "do you know how to code?" not on the list?

Because AI writes that part. You describe the game in plain words, and AI turns that into the commands the computer understands. If you want to know how that works today, see whether you can make a game without knowing how to code.

That removes one obstacle, but it does not remove the work. AI builds what you ask for; the one who decides what to ask, plays, notices what got boring, and asks again is you. The five questions below are about that part, the one that stays with whoever makes the game.

Question 1: do you want to understand why a game hooks people?

If the answer is yes, you already have the fuel that lasts longest. People who are curious about why a game is fun have plenty to think about for a long time, and that shows up in every decision: what the player does, why they want to play again, what to cut (our reading).

Real cases tend to start this way, with a taste for something, not a business plan. Jonathan Bjerk, an accountant in Norway, grew up playing a snake-style game with friends, all sharing the same keyboard. According to him, Curve Clash, the game he made with AI, came from wanting to bring that same feeling to phones and to players around the world². The full story is in Curve Clash: the accountant who launched a game made with AI.

Masako Wakamiya, a retired bank employee in Japan, started making her game at around 81, without AI. Her reason, according to Gigazine's report, was simple: losing to young people at games was no fun, and she wanted a game where people her age could win³.

What to do about it: if you cannot say which game you like and why, start there, playing with attention, before you think about making one.

Question 2: can you test the same thing many times?

You need to, because AI makes mistakes and you are the one who finds them. In an academic test presented at ICML 2026, an AI research conference, GameDevBench, the best AI setup solved 53.8% of 333 game tasks on its own¹. The rest depends on someone playing, pointing out what broke, and asking again.

The test itself shows where AI stumbles most: on 2D game art tasks, success dropped to 33%, versus 51.4% on gameplay tasks¹. It is a lab measurement, with tasks taken from tutorials, and it does not tell you how much will work in your game. But it gives you a sense of how much checking there is to do.

In practice, that turns into repetition. Bjerk says he spent hours just playtesting and tweaking numbers, and that he was stuck on bugs in the multiplayer mode for about two months, in September and October 2025². What is still hard with AI is covered in is making a game with AI easy?.

What to do about it: if testing the same level ten times sounds unbearable, know that it is a big part of the work, and pick a first game where each test is short.

Question 3: do you have a little time almost every day?

Twenty minutes a day already counts, in our reading. A little time, often, tends to get you further than waiting for the free month that never comes, because the game stays fresh in your head and each session picks up where the last one stopped. This is a routine suggestion, not a deadline.

Real cases show people who made games in their leftover hours. Bjerk still works full time as an accountant. He started the game on May 28, 2025, opened a test version in August, and launched the full game in mid-December, less than seven months later, according to the interview². That was his time, with a small game.

Before AI, it usually took much longer. Stardew Valley, a farming game, took four years of work by one person, Eric Barone, according to the game's official page⁴. Masako Wakamiya took more than half a year to make hers, studying from books and asking a teacher questions over video calls³.

What to do about it: look at your week and mark where twenty minutes would fit. If they do not fit anywhere right now, the honest answer is "not yet", and that is fine.

Question 4: will you start with a small game?

You need to, because your dream game is usually too big for a first game. A small game has one action, one screen, and short rounds, and that is why you can finish it, test it, and show it. The big one comes later, once you know what the work is like (our reading).

Games that became famous started this way, and none of them was made with AI. The first version of Flappy Bird, the game where a little bird flies between pipes, was made in two to three days by Dong Nguyen, who had been coding since he was 16⁵. Minecraft had its first public version in May 2009 and was only considered finished in November 2011, more than two years later⁶. Its creator, Markus Persson, already worked as a programmer⁷.

These are cases that worked out among many that did not. What they show is the size of the start, not the chance of success. To trim your idea until it fits, see the first step before opening any program.

Question 5: do you have the nerve to show a flawed game?

You will need it, because you only find out if a game is fun by watching someone else play it. And the store requires it from beginners. On Google Play, the Android store, a personal account created after November 13, 2023 can only launch after a closed test with at least 12 people, for 14 days in a row⁸. When you apply to launch, the store asks for a summary of what those people said and how you collected their feedback⁸.

In other words, before it reaches any stranger, the game goes through twelve people who will find flaws. The store's own page recommends keeping a record of the feedback, replying to testers, and fixing bugs during the test⁸.

Those who did it before showed their games early. Minecraft was opened to the public while still unfinished, in 2009, and Persson kept updating the game based on what the community said⁶. Bjerk opened the Curve Clash test version four months before the full launch and, according to him, his advice is to playtest with real people early and often².

What to do about it: think right now of two or three people you would show an ugly, unfinished game to. If no one comes to mind, start looking before the game exists.

What if your only reason is to make money?

It is worth knowing, before you start, that most games earn little. According to an analysis based on Gamalytic, a company that estimates sales, about 65.9% of the games released in 2025 on Steam, the largest store for computer games, earned less than $1,000⁹. These are estimates, not official store data⁹.

That does not say no one earns money, or that your game will end up in that group. The point is different: money alone is a weak reason to get through questions 2 to 5, because the return, when it comes, takes time and has no guarantee (our reading). No one can promise you will get rich. As in any profession, there is risk.

If your honest answer to question 1 was "I do not care about games, I just want income", maybe the path is a different one. If it was "I like games and would also like to earn from them", the five questions still apply.

The other side

The five questions are our reading, not a scientific test. Answering yes to all five does not guarantee you will finish a game, that anyone will play it, or that it will earn money. Answering no to one of them does not mean making games is not for you either: sometimes it means "not yet" or "at a different size". The cases cited are few, and they are from people who made it to the end; most people who start never show up in a news story. And Curve Clash, Hinadan, and the famous games each had their own story: none of them is a deadline for anyone.

What if I answered "no" to one of the questions?

It depends on which one. A "no" to question 1 is the most serious, because taste is what holds up the others. A "no" to questions 3, 4, or 5 can usually be adjusted: less time calls for a smaller game, and the nerve to show it grows by showing it to one person first (our reading).

Question 2 sits in the middle. If testing many times sounds boring, make a game where each round lasts seconds: the test gets short and the repetition weighs less.

And it is worth looking at the cost side before any decision: you can answer the five questions without spending anything, and only then see how much it costs to make a game with AI, item by item.

What should you do before spending a dime?

Answer the five questions in writing and test the answers in real life, without paying anything. Playing with attention, marking your free time in the week, and thinking about who would see your game are things you can do today.

Your action plan

Check off what you've done. It's saved on this device, so you can pick up where you left off.

1. Write the five questions on a sheet of paper and answer each one with yes, no, or sort of.
2. Write down the name of a game you like and one sentence about why it hooks you.
3. Mark in your week where twenty minutes would fit almost every day.
4. Write the idea for your first game in three sentences and cut it until it fits on one screen.
5. List two or three people you would show the game to while it still has flaws.
6. Reread your answers and circle the question that weighs most for you.
7. Hold off on paying for the AI and the store until you have the answers.