4 ms·

Man the one-shot game that is genuinely good in 2027 is crazy. A good game typically takes around 3-8 years of development by multiple skilled people. Maybe 3 y

Man the one-shot game that is genuinely good in 2027 is crazy. A good game typically takes around 3-8 years of development by multiple skilled people. Maybe 3 years or so for 1 guy that's super dedicated. Right now single prompt with Fable can get us a small protype of like 1 game mechanic that's not even remotely production ready. So this guy thinks it'll 100,000X in a year. AI boosters are something else.

inigyou 3mo ago
It's like any time I ever wanted to make a game. "What if there was a game where you could drive around and shoot people?" Oh it turns out driving a square to throw circles at other squares isn't actually very fun. AI does the same.
anuramat 3mo ago
> 100000x in a year why not? fable 5 (2026) is an inf% improvement over having nothing (2022) in any case, stop pretending that it can be represented by a single scalar value, or at least use log scale units
- jenniferhooley 3mo ago
  Oh, you think we are going to have another breakthrough in artificial intelligence rivaling the one we had in 2022 this year or next? What research or studies lead you to believe this? Could I see one? I haven't personally seen the research leading me to believe there's a massive new intelligence breakthrough on the horizon but maybe I missed it! As to the current version - I use AI constantly, as in 12 hours a day 7 days a week specifically for high-end game development. It is currently so far from the 2027 prediction that it's hilarious if you think a little incremental improvement (e.g., like from opus 4.6 to Fable) is going to bring us there.
  - anuramat 3mo ago
    > this year or next maybe; I meant "in a single year eventually", not this year necessarily > what research transformers were published in 2017, and nobody anticipated agentic coding back then; but if I had to guess: 1. lean self play as in https://arxiv.org/abs/2502.00212 https://arxiv.org/abs/2502.00212 and https://arxiv.org/abs/2606.13473 https://arxiv.org/abs/2606.13473 2. "recursive self improvement"/autoresearch even basic stuff like coding agents removes a major bottleneck -- when finding relevant literature and turning an idea into code takes as much time as writing it down in your notes, researchers from top tier labs can test every single hypothesis they can come up with > hilarious "oneshot a PoC -> oneshot a production ready game" improvement sounds like something that will eventually be done in a 1kloc diff in the RL pipeline IMHO: "genuinely good" might never happen though because of all the non-coding stuff -- you can't really benchmaxx when the target is defined as "pleasant to humans"

inigyou 3mo ago

It's like any time I ever wanted to make a game. "What if there was a game where you could drive around and shoot people?" Oh it turns out driving a square to throw circles at other squares isn't actually very fun. AI does the same.

anuramat 3mo ago

> 100000x in a year why not? fable 5 (2026) is an inf% improvement over having nothing (2022) in any case, stop pretending that it can be represented by a single scalar value, or at least use log scale units

jenniferhooley 3mo ago
Oh, you think we are going to have another breakthrough in artificial intelligence rivaling the one we had in 2022 this year or next? What research or studies lead you to believe this? Could I see one? I haven't personally seen the research leading me to believe there's a massive new intelligence breakthrough on the horizon but maybe I missed it! As to the current version - I use AI constantly, as in 12 hours a day 7 days a week specifically for high-end game development. It is currently so far from the 2027 prediction that it's hilarious if you think a little incremental improvement (e.g., like from opus 4.6 to Fable) is going to bring us there.
- anuramat 3mo ago
  > this year or next maybe; I meant "in a single year eventually", not this year necessarily > what research transformers were published in 2017, and nobody anticipated agentic coding back then; but if I had to guess: 1. lean self play as in https://arxiv.org/abs/2502.00212 https://arxiv.org/abs/2502.00212 and https://arxiv.org/abs/2606.13473 https://arxiv.org/abs/2606.13473 2. "recursive self improvement"/autoresearch even basic stuff like coding agents removes a major bottleneck -- when finding relevant literature and turning an idea into code takes as much time as writing it down in your notes, researchers from top tier labs can test every single hypothesis they can come up with > hilarious "oneshot a PoC -> oneshot a production ready game" improvement sounds like something that will eventually be done in a 1kloc diff in the RL pipeline IMHO: "genuinely good" might never happen though because of all the non-coding stuff -- you can't really benchmaxx when the target is defined as "pleasant to humans"