Is building games faster really better?

General
1

This question has been prompted by the ongoing AI discussions on these forums and elsewhere. When users of this tech are asked why they use it, I’ve noticed they often state that it helps them develop games faster or more efficiently. The question I’m asking is whether this is actually a benefit.

I’ve made comments before talking about how the more prolific AI game creators I’ve seen don’t appear to be very successful, but I never dove into the data to find out if that was an industry-wide trend, or isolated to just the most egregious examples of “AI slop”.

Pulling data from Steamdb’s releases page we can see a sharp increase in the overall number of released games in 2024 and 2025 compared to previous years. However, this increase seems localized to what Valve considers “limited games”.

image
image1016×867 48.8 KB

I single out 2024 and 2025 specifically as these are the two full years most affected by the recent AI boom, and I think it’s fair to attribute the increase in released games to the increased “speed and efficiency” of AI-assisted game development. From 2019-2023 (the pre-AI Steam Direct era), we saw a year-on-year increase in the number of released games in the approximate range of 1,000-1,800 games/year. In the past two years, we’ve seen a growth of 4,487 games in 2024, and 2,921 games in 2025; we are almost halfway through 2026 and by my estimation we’re on track for a growth of 2,672 games by year end.

Naturally, reviews in all brackets have seen more growth in 2024-2025 than prior years, but reviews in the 0-99 range seem to have increased more per-capita than reviews in the 100-500+ range. This suggests that while some AI games are getting favorable numbers of reviews, the majority are getting 99 or less, with a less-favorable ratio than non-AI games.

To simplify the math involved, I’ll take a smaller sample of the data:

  • In 2023, 14,045 games were released total, 4,336 had reviews in the 1-9 range, and 1,167 in the 500+ range.
  • In 2024, 18,532 games were released total, 6,212 had reviews in the 1-9 range, and 1,391 in the 500+ range.
  • That’s a 31.9% increase to total games released,
  • A 43.2% increase to games with 1-9 reviews,
  • A 19.1% increase to games with 500+ reviews.

Again, this is just a snapshot, the full data on the website linked at the top shows the rest. Also, apologies if the math isn’t 100% right, I’ve checked it multiple times, but I was never taught statistical math.

This snapshot correlates with my observation that games released during this “AI era” are getting worse reviews on average. I don’t have the data to check if this correlates with worse sales, for purposes of this post I’m just going to assume it does. If this trend is caused the performance of AI-developed games, then it suggests that any advantages gained from the “speed and efficiency” of AI game development do not translate to better performance on Steam.

Now, I made this post just to share the data and my inferences made from it, so I’m not going to speculate on why the AI-developed games may be performing worse than traditionally-developed games, but I’m happy to discuss that in the comments.

Also, if anyone’s got suggestions on how I could display this data better, please let me know.

2

Sorry to disagree but I am a bit sceptical about this claim and would need to see more research and evidence to believe that AI is the cause of that. There could be and probably are many more reasons and my guess would be that AI is not a major factor in this at all.

4 Likes
3

Here is a Google Trends for youtube searches for ‘godot’ in the UK.

image
image1534×390 34.1 KB

Is that blip in 2023 to 2024 also AI?

Correlation is not causation.

Although it is a bit of a downer to see how many games are being released. Thanks for sharing that data.

1 Like
4

My logic was the AI boom hit at the same time as an extra 3,000 games over average growth. I’m sorry if I was jumping the gun thinking they were related.

5

Well you might be right. I am just not convinced yet.

6

That first blip there in 2023 was probably 4.0 being released on March 1st, and the second is almost certainly the release of 4.1 in July of 2023

4 Likes
7

Fair. If I had access to the data for how many games are tagged with AI disclosures, that may be more convincing. I think that data exists on the paid games data sites, like VGInsights, but I don’t have access to the paid part of their data. I also don’t like the way they present the free data on most of those sites, seems ripe for misinterpretation.

8

I think that assuming that “AI” drives this is giving it far too much credit, the shift is probably simply that availability of indie tools have significantly increased in this period, and probably other factors making people invest more time and effort into making games, with the economy being what it is it’s a more and more attractive solution to not being able to get employed I’d say

And honestly if it is really an increase starting in 2023 the release of Godot 4.0 is not impossible as an actual cause of that boom, being a royalty free alternative to the established engines, and being released in 2023, and becoming truly competent and stable in 2024 with 4.3

4 Likes
9

I’d also argue that if we’re including 2024 in the question “AI” is very unlikely to be a major factor as it was pretty marginal until last year

1 Like
10

Though, would that also explain the decrease in performance of games per-capita? Also, my findings show a 31% increase in released games on Steam in 2024, meanwhile the largest estimates I’ve seen for Godot’s industry adoption is 10% during that same year.

11

So I don’t think Godot’s 4.1 release is really the answer either, unless Godot users are making more than one game per year on average.

12

Where did I say the release of 4.1? Or are you mixing up my reply to the graph of searches for “Godot”?

Also what portion of the influx of games are “industry” rather than indie who might not be reflected in such stats

But given how different the adoption and availability of “AI” was in 2024 I doubt it would have been any more likely to be a driving force then

13

Also for reference this is the games published using Godot over the last 8 years:

1000016734
1000016734517×371 9.86 KB

Not the entire equation but a good representation of the indie portion and influx

14

This is itch.io which is a much greater number, not sure how representative the Steam side is as it depends on their API and self reporting

1000016735
1000016735600×371 10 KB

15

I think I’m getting mixed up, sorry.

I can’t find the exact report I was quoting there, I think VGInsights had it on their blog, but they took down their blog. I’ve found a survey from GDC that same year that shows a 3% adoption rate, granted they’re geared towards professionals, when the other report I read included hobbyists.

image
image1091×847 43.6 KB

Link to access report: GDC 2024 State of the Game Industry Report

1 Like
16

Anyway, I’ve got to go, it’s 11pm, I’m running on 4 hours of sleep and got a meeting in the morning.

1 Like
17

Therenis no evidence, theres a correlation, but no causal link. You could guess that its one of these …

  1. Coincidence
  2. increased Godot adoption
  3. AI help (a) LLM, (b) art generators
  4. Developers cashing in on games theyve been sitting on before its too late.
  5. Steam is growing
  6. Developers fleeing normal work like websites and database code and escaping into games.

Theres a lot more … but I think AI can be considered part of overall growth too.

3 Likes
18

i think the amount of games is connected to something i’ve personally experienced - which is avaibility & popularity of tutorials

before 2020 tutorials were few times rarer, while game dev was more of a niche than something mainstream / indie

also (i can speak for non-english speaking country) thanks to rapid digitalization in the last decade many people from non-english countries managed to learn english faster, from gaming, from new culture ect. thus increasing number of sources from which to learn

you can see the growth is attributed from 2019, not from 2023, and it’s slighty expontetial in 2023, probably due to Ai, but i’d argue that the invisible foundation had bigger impact than bad AIs from last years, AI became good only about year ago maybe, before it was terrible to make games

also - remember to take this from standpoint of non-dev at that time, cuz if you were already in game dev you probably seen no difference in number of tutorials

1 Like
19

Coding is maybe like Painting: You have a small precision tool and slowly generate a big picture from it.

Writing Code with the AI can be like many things but if done remotely well it is more like stonework:
You have a rough tool with some control over what you want to do but only little fine control ( unless you use specialized tools ) so the AI decides a load and you have to ensure it makes the right decisions.

Now is fast coding better?
That really depends if you take the time for the details and have finer tools too: Reworking AI code, Human review, implementing little detailed features, using the AI to add details can also speed it up but it does not mean it is bad, it just usually is because most people have no eye for details anymore and using AIs trains People to get even less caring about details.

Compare it to fast food:
Fast food can be healthy, its just prepared food you can eat quickly, it can be healthy if it is prepared properly, but usually it is not.
The big problem is: If People accept fast food food companys want to maximize profit and the food will likely be unhealthy.

AI is the computer fast food but it is not something bad, it is just a tool that can be used in ways that are ungood.