from Hacker News

OpenAI, Stanford study finds LLMs can boost public opinion manipulation

by mariambarouma on 1/14/23, 7:10 PM with 1 comments

  • by 082349872349872 on 1/14/23, 7:31 PM

    There was an article recently that it had become very difficult to provide official cover, as social media backstories are not as easy to populate as official backstories.

    LLM-generated cliques might both solve that problem and (in conjunction with spearphishing) address the "bootstrap" problem of flooding cliques that don't have organic neighbours?

    > even the most advanced LLMs tend to make absurd statements and lose their coherence as their text gets longer than a few pages.

    If not making absurd statements and remaining coherent over pages is what it takes to pass a Turing test these days, we'll see failures for many wetware content generators.