The talk about hallucinations in LLMs has gotten it all wrong. The true hallucinations are by company execs who think it is OK to release to general users products that are based on LLMs that confabulate wildly, as all LLMs do. Time will show a high price paid by society.

Amazon Fresh failure. My third law of robotics, shared here yesterday is: 3. Technologies for robots need 10+ years of steady improvement beyond lab demos of the target tasks to mature to low cost and to have their limitations characterized well enough that they can deliver 99.9% of the time. Every 10 more years gets another 9 in reliability. Looks to me like Amazon did not remotely follow this when they introduced Just Walk Out to their Amazon Fresh stores in 2016. Touted as cameras seeing what you put in your cart there was no need to check out. Service cancelled this week. Turns out there were 1,000 remote cashiers in India: "It often took hours for customers to receive receipts after leaving the store, largely because offshore cashiers were rewatching videos and assigning items to different customers." This story reports that in 2022 only 300 out of 1,000 walk outs did not need a human review, so 30% reliability against an Amazon target of 95%. Their business model built in a 5% failure rate, and even that was not even close to attainable. Not the magic of AI that consumers believed they were partaking in. Hubris gets you every time in robotics (and AI...). gizmodo.com/amazon-reportedl…