@nicochristie one could say beyond a certain threshold of difficulty, latency stops mattering coz you're like take a full hour but solve this problem? arguably, mythos-level intelligence is only needed for tasks at that level of difficulty. wdyt?
@rauchg so true, you need to be a software engineer to ensure AI agents can vibe code sustainably. What's a good metric to measure judgement for engineers and projects?
@jxnlco any tips on updating your AGENTS.md (et al) so that codex learns your preferences over time? so far, I ask codex any learnings you'd like to add to agents.md after a long session. but I want something more automated and continuous.
@nikitabier Please add reading time estimate to X articles. Over time, it should be personalized to a user's reading speed and propensity to quickly scroll the genre.
@nikunj As you alluded in "explore all edges" - exploratory budget should be pretty high but prod budget needs to be capped. Mixing the two leads to unjustified spend.
the input interface has been the same for decades.
with ai, software can now reason and act on your behalf but the interface is the bottleneck.
why do i have to check sushi on 10 restaurants across 3 apps? why can't i just do it with a flick of a finger?!
the world's about to get a new interface @agi_interfaces
@nikunj the perf difference b/w models on benchmarks is not high enough to justify switching costs (claude.md, muscle memory, skills, etc.) for me. If and when it becomes high, I'd definitely try a new model.
@andrewchen spreadsheets are easy to verify. that made mistakes easy to catch. as spreadsheet users become vibe-coders without the skill to verify the generated code, there will be a plethora of undetected errors.