[ad_1]
MORGAN PARK, STAFF WRITER
Last week: Finally finished my replay of Metal Gear Solid 3 and decided it’s easily the best one.
Yesterday, a coworker shared a link to a little browser game that looked right up PC Gamer’s alley: Doomscroll, a top-down shooter played entirely with a scroll wheel. You move your little Doom-guy facsimile down a hall, automatically shooting forward while dodging flanking sprites. Occasionally, you pass by plaques etched with actual New York Times headlines from the current day, completing the doomscrolling wordplay.
It was kinda fun: I played a few runs, unlocked some upgrades, and experienced brief whiplash after locking eyes with two headlines about the assassination of Charlie Kirk. After, I clicked over to a newsletter the developer wrote about making the game. Doomscroll is a side project of filmmaker and writer David Friedman, who describes in detail how he generated the entire game using ChatGPT.
“As readers know, I’m not a coder, but I enjoy how vibe coding lets me turn an idea into something real. So naturally, I turned to vibe coding for this,” Friedman wrote.
Friedman goes on to describe his process of telling the chatbot to make reverse Galaga with automatic shooting, basic upgrades, some obstacles on the path, and a wall fire that kills you if you move too slow. Ah, that familiar feeling of deflation as the charming qualities of a thing vanish before your eyes.
If there’s anything the rise of generative AI has taught me, it’s that I’ve taken the gift of human authorship for granted. Until recently, I’ve had the privilege of knowing that any game I play began as nothing, and became something through choices, collaboration, creative problem solving, and picking up a (digital) paintbrush.
I can now acknowledge that one of the best reasons to play games is to be in conversation with the people who made it, even if those conversations have prickly questions like “Why the heck did they make this boss so hard?” or “Are you sure about these guns?“
Playing Doomscroll did inspire questions: Why do blocks and spider webs spawn in places that can’t possibly hinder me? Why are the headline plaques so awkward to read? Why do all the monsters look like they’re in a different game than the main character? Unfortunately, I suspect the answer to all of them is “because the AI spit it out that way.”
When I assumed Doomscroll was the work of a trained game dev with some free time, the ways it was sorta crap didn’t bother me in the slightest, but heavy use of AI generation has a way of erasing the vulnerability of whatever it makes up. Funnily enough, the newsletter shares a screenshot of a Doomscroll prototype that featured hand-drawn pixelated monsters.
“Next, I set about making pre-rendered monsters. But wow, small-scale pixel art is hard. I went through a lot of versions of different monsters. Eventually, I had a few I felt were good enough,” he wrote.
They looked kinda cool, and had more personality than the final product: a gaping maw beast with long teeth, a floating skull with green fire, and a tentacled eyeball with an unimpressed eyebrow. In the end, Friedman cut them in favor of generic AI monsters that look laughably bad by comparison.
“It had its own charm, but in the end, I didn’t love it. Somehow, my pre-rendered monsters made the game feel worse. Maybe it’s because I just am not a good pixel artist.”
On that last point, we’ll have to disagree. By my estimation, the most human thing about Doomscroll is that double entendre title. Friedman gives some lip service to the metaphor at the heart of the game at the beginning of his newsletter, saying that people spend “too much time scrolling endless feeds of content that make you feel bad about everything.”
All too relatable. Though, as the dangers of generative AI are increasingly the subject of such doom scrolls, I wonder if Friedman intended some irony. As a free game that required just a few hours of chat bot fiddling over Friedman’s recent vacation, Doomscroll is an inoffensive exercise, and the accompanying blog has some interesting insights into what it’s like to essentially argue with a robot until you’re satisfied (or happy enough) with its output. Sounds frustrating.
[ad_2]
morgan.park@futurenet.com (Morgan Park)
Source link