Idk if someone paying attention to how 2025 and 2026 have gone thinks that by 2028 we will be backing off of agenting coding that is wild. Like the other comment says: future models refactor the code of older models.
You can use LLMs heavily without ever actually "vibe coding". I do think to the degree "vibe coding" continues to exist there will always be work to do in turning some portion of vibe coded work into more robust production quality code. You can still use LLMs to do this you just have to maintain control over architectural choices.
Yea it’s hard for me to think of what the end state equilibrium is. A pile of vibe coded junk today is bad. You need humans. But we’ve made such a ridiculous amount of progress in such a short amount of time and, most importantly, this shows no sign of slowing down or plateauing. So will we hit a point where “vibe coding” is just all there is? Where human intervention is bad, just as hand tuning assembly is bad?
Is there a level of abstraction where human involvement will always be necessary? If so where?
I think it's not going to be a straight line upwards, it's going to get weirder. LLMs are still riffing on what we humans have done. If we stop giving them good examples because we aren't architecting anymore I don't really know what's going to happen. Probably some kind of meta architecture that emerges from lots of agents working in parallel but I worry it might have a strong house of cards theme to it.
if you were paying attention you would've noticed that between 2025 and 2026 the pricing of these things have somewhat changed. How does the extrapolation look with that?
Comments
Idk if someone paying attention to how 2025 and 2026 have gone thinks that by 2028 we will be backing off of agenting coding that is wild. Like the other comment says: future models refactor the code of older models.
You can use LLMs heavily without ever actually "vibe coding". I do think to the degree "vibe coding" continues to exist there will always be work to do in turning some portion of vibe coded work into more robust production quality code. You can still use LLMs to do this you just have to maintain control over architectural choices.
Yea it’s hard for me to think of what the end state equilibrium is. A pile of vibe coded junk today is bad. You need humans. But we’ve made such a ridiculous amount of progress in such a short amount of time and, most importantly, this shows no sign of slowing down or plateauing. So will we hit a point where “vibe coding” is just all there is? Where human intervention is bad, just as hand tuning assembly is bad?
Is there a level of abstraction where human involvement will always be necessary? If so where?
I think it's not going to be a straight line upwards, it's going to get weirder. LLMs are still riffing on what we humans have done. If we stop giving them good examples because we aren't architecting anymore I don't really know what's going to happen. Probably some kind of meta architecture that emerges from lots of agents working in parallel but I worry it might have a strong house of cards theme to it.
Great that you know the future when nobody else can.
No, I just have a prior that’s informed by real information and evidence and that’s quite strong.
- scaling laws exist
- downstream perf trends also exist (epoch capability index)
- gpt4 to gpt5 leap in capabilities every 16-18 months
- actual adoption and retention and engagement numbers are out of this world
- RL with verifiable rewards will get you to super human performance even with poor sample efficiency
- zero evidence of a plateau
I don’t know the future I am just skeptically looking at the data.
What data are you looking at?
if you were paying attention you would've noticed that between 2025 and 2026 the pricing of these things have somewhat changed. How does the extrapolation look with that?
Good thing we have reams of data on this, holding performance constant the cost goes down 10-40x per year: https://epoch.ai (like the first box)
Also, frontier token prices have remained roughly constant:
3.5 sonnet: $3/$15 3.7 sonnet: $3/$15 Opus 4: $15/$75 (opus tier) opus 4.1: same Opus 4.5: $5/$25 Opus 4.6 (same) 4.7 (same) 4.8 (Same) Fable: $10/$50
So Fable is cheaper than Opus 4 was at launch.
One thing that has increased quite significantly? Spending and adoption.