@felixrieseberg@tszzl i replaced coding interviews at my company with AI building interviews and the signal from them is way more useful. biggest surprise is how easy it is to screen for ambition.
@evergreencap3 imo commodification is less likely for high alpha tasks because the performance is gated by having access to quality proprietary data specific to that task/industry. labs are pursuing vertical plays by buying businesses producing this data. moat unless you steal the data.
what happens when a normie just accepts it as a totally normal thing that they can one shot a saas with a stripe checkout without having to even know what stripe is? imo we’ll be there quite soon and it will be cheap to execute that
imo we’re heading towards a world of on demand individualized software for anybody: just like today enthusiasts can conjure up interactive explainers and simulators using frontier models, this will become a standard verb in the hands of any generic end consumer of LLM products
example: it takes 5 minutes for opus 5.5 to make an interactive DAW that teaches you music production step by step. this would unthinkable two years ago and normies aren’t thinking like that yet.
what if we scale it up? spin up a whole AWS based service with heavy lifting infra
tech bois at sf parties talking about their personal lives:
▶ 🔘─────── 00:02
tech bois at sf parties talking about their startup:
▶ 🔘─────── 38:07
tech bois at sf parties talking about the WeWork IPO:
▶ 🔘──────── 2:58:47
Excited to announce our recent work on generating 8-bit music using Transformer! Our LakhNES model uses transfer learning: we pre-train on the heterogeneous Lakh MIDI dataset and fine-tune on NES music
🎶https://t.co/1XrBgzJFz3
📜https://t.co/idRGXU5Vef
⭐️https://t.co/vgHkYkGNEH