i wrote precisely this stuff as soon as i landed here 2 years ago and saw everyone was claiming LLMs were conscious, clearly motivated by the fact that they were talking to them. got basically no reply
1 more tweet in this thread
disclaimer: i'm not agreeing with all chiang says. clearly. the claim below makes no sense and there *are* arguments for why gpt-4 is more likely to be conscious than alphafold. i just never read them here. i only saw strong bias towards the specific thing the net was trained on
this is an interesting point in the new ted chiang piece – no one really claims that alphafold is conscious, or that sora or midjourney or dall-e are conscious
retweets 0
likes 13
substance 5 · argument · rank 75 · People claim LLMs are conscious largely due to conversational bias rather than sound arguments.
view on X ↗