@kbc

All AI sound the same...But it's not just that they sound the same, but they also sound very different to humans. Essays written by AI (green crosses) are semantically different from human essays. That doesn't just mean they use different words. It also means they use similar arguments and draw similar conclusions. For AI to be helpful for humanity, it's thinking should be all over the places. We need divergence in thinking, weakly connected clusters of opinions, decentralized decision-making. Call it what you want, embedding in it whatever philosophy is closest to *your* heart. Source: How LLM distorts our written language https://arxiv.org/pdf/2603.18161v1

0 reply

0 recast

2 reactions