I use AI tools every day, and yet Worth a Watch leaves out AI-made releases. You deserve a full explanation. Am I a hypocrite?
This AI debate keeps coming up, in many cases devoid of nuance or the required context. I thought I would put something in writing that I could refer people to because 240 character responses are seldom enough, and I’m sure you have witnessed these discussions get heated yourself. For good reason!
The retro scene is, of course, made up of humans with a wide variety of positions, but the general conversation keeps coalescing into two warring camps:
- People proudly using AI tools, some to avoid hand-coding.
- People who are against AI being used at all.
Where’s the line?
As things progress, it is becoming an effort to avoid AI in our day-to-day computing lives. Google search defaults to it, Microsoft keeps adding it to where it is not welcome or needed, and many employers are either forcing it or strongly expecting it.
On the other hand, there is a huge “ick” feeling for a lot of us when the AI has been used carelessly, tastelessly and obviously.
Right now, based on this temperature check of the community, Worth a Watch leaves out “AI-made” releases (games, ports/decomps, FPGA cores). I define “AI-made” as versions where a model did the creative or engineering work.
But other than the “I know it when I see it” worst-case scenarios, i.e. six-fingered “AI slop”, and unless you are in the “no AI ever” camp, we need to be careful about where we draw the line.
How much AI assistance is OK? Is it defined by which part of the workflow? Is it defined by how it is used?
Where I stand (today)
I change my mind based on new data and experiences. That said, I have a fairly well-considered opinion right now. It might upset some of you, and I fully appreciate that.
I’m not anti-AI.
In a LinkedIn comment, around June 2026, I said the following:
“If you tell a computer in great detail how to do something that is programming whether you used English, C++, Python or BASIC. ‘Gimme a game plz‘ isn’t programming, but thinking through the end goal and working out the best way of achieving it while taking into account […] is.”
What I AM against is slop. Low-effort output, no human in the loop, abdication of responsibility, effort, or creative input.
I’m also against the damage AI companies are doing to the environment, to communities, to economies, to individuals, and the wider human race.
As a traditionally published book author, I personally feel anger at our protected intellectual property being hoovered up at industrial scale without any compensation or even permission.
I am not as pure as the anti-AI guy, though. Despite my obvious reservations, and constant mockery of AI-maximalists and the AI financial bubble, I do use AI tools (as I type this, I have both Cursor IDE and Claude subscriptions).
Programmers are likely the most common kind of folks who use it in their work with some definable ROI (studies vary).
It’s helped me a great deal in “programmer jobs that aren’t the actual coding” tasks such as GitHub automations, docker setups, backup schedules, debugging, bulk-transcriptions of my workshop recordings, finding stuff I have lost, that kind of deal.
Yeah, nothing I can’t do without AI, but the tools do it more quickly and with less Googling. That includes Worth a Watch itself as it helps me filter the fire hose of RSS feeds, community posts, Reddits, YouTube videos etc and categorise them. It’s hand-curated, I pick every story and edit what goes out, but with out that automated assistance it definitely couldn’t be weekly.
Yeah, these tools are a part of my work today. That stated, I am always in the loop and would be scared actually to NOT be. I’ve seen what can happen when they are untethered, but also I am witnessing daily people relying on them instead of their own brains.
On that note, I am ashamed to say I am much less likely to get a regex correct first try now. I rely on Cursor so much for that. Brain rot is real.
So, TL;DR
Effort and human review are the line I draw, I think, not the tool in the mix. Grammarly was involved in the editing of this post, but it is still me literally typing these words.
Detailed instructions, steering the progress toward a clear end goal, and checking/validating the results along the way. That is programming in my view, whatever the language being used.
“Make me a game where Mario races Sonic” is not programming any more than selecting a specific espresso-based beverage at the coffee shop makes you a barista.
Should Worth a Watch include AI-assisted work?
With the above in mind, how do we judge a project on its merits, not as a ruling on the tools used?
There are a lot of hackathons, scene parties, and game jams now scrambling to come up with fair rules due to AI submissions. Being flooded with generated code has caused more than one open source project to stop taking pull requests, and bug bounties such as curl’s are closing for the same reason. The sheer quantity is only going to go up.
It’s not an entirely new problem, back in the day competitions would exclude games made with the Shoot-‘Em-Up Construction Kit and Graphic Adventure Creator, and even STOS and AMOS work. The thinking being those tools made things too easy.
So for now my answer is no: Fully AI-made releases stay out, judged on whether an actual person did the majority of the creative or engineering work, not on whether a tool was involved somewhere along the way. I will keep revisiting that as things change.
What RGC is about either way
Regardless of what you use in your toolbox, my goal is to help more people become happy and fulfilled retro coders. To feel like you understand more than when you arrived, and are equipped to manifest into reality what you only currently imagine.
I would say, if you are going to use AI, ask it to explain to you rather than do it for you, and understand as much as you can so you can at least tell when the answer is wrong.
If you do want the week’s retro news with a human doing the choosing, Worth a Watch comes out every week.
Original article by retrogamecoders.com
