@magofoco
The old 20% is the new 80%.
LLMs keep dropping every month.
OpenAI → Anthropic → Google → Grok → repeat.
Scaling is basically done with the current architecture. Models are hitting the ceiling.
Claude Opus 4.5 just hit 80% on SWE benchmarks. Humans are shaking.
What’s left?
The cursed 20%.
Judgment, taste, context, not being an NPC (very hard)
Invest there if you do not want to get replaced.
The first 80% is already gone.
0 reply
1 recast
1 reaction