Summary: If widely and successfully deployed, LLMs will eventually start digesting each others output and form one big, gigantic feedback loop that ultimately t

Summary: If widely and successfully deployed, LLMs will eventually start digesting each others output and form one big, gigantic feedback loop that ultimately tends toward what ... most likely random nonsense and incoherence ... that may or may not sound plausible. Conclusion: LLMs are still as dumb as a hammer because they can't identify their own output as the bullshit it is. This is *not* real AGI but rather just a poor facsimile of it. The only way these bullshit generators can destroy humanity is if humanity itself is bullshit.