Why Almost Nothing You Read Online About AI Is True

I’ve adapted my Special Theory of Autonomous Agent Reliability (STAAR) to predict the reliability of social media* posts about “AI”. In STAAR, the reliability of a single step – a single model interaction – is given as: R = 1 – (1 – C)(1 – P) Where C is the probability the model’s output is correct, and P is the probability that any errors are caught before they propagate and compound. So the reliability of N steps in a workflow is Rᴺ. (It’s essentially Lusser’s Law of system reliability applied to processes). We see a similar thing with social media AI content. Influencer A writes a post about AI. It has a probability C of being accurate. Influencer B reads that post with a probability P of spotting any errors or untruths before they write their own post propagating and compounding those errors. The odds of reading anything accurate about “AI” on social media that has propagated through its Small-World networks are low. I call this my theory of Social Transmission Accumulated…