I am reminded of using route finding mapping programs in the early 2000s. Once one took me 15 miles through the desert to put me on a road with an odd name something like "Overland Stage Road" or something. When I got to the turn I was supposed to make, I saw a locked gate with a narrow, over-grown dirt road behind it. Apparently the database the mapping program was using was wrong about the public accessibility and quality of the Overland Stage Road and I had to backtrack 15 miles to get to where I wanted to go. Another time, routing to my brother-in-law's house in Houston, the mapping program brought me to the main boulevard just past his house and had me drive up through the parking lots behind some medical buildings to park at an 8 foot tall concrete wall that separated my brother-in-law's neighborhood from the medical complex. If he had been in his back yard he could have heard me screaming.

Remarkably, years later, it has been decades since I have experienced an error comparable to that from a mapping program.

You are using a product that seems like a customer service human and assuming that it is more like a customer service human than it actually is. As currently implemented, there is no feedback path between your "bug report" such as it was, and anything that is going to be directly used to fix the product. And yet our experience, or at least my experience, with mapping programs suggests that mistakes such as the one you have found and described so completely will decline immensely in rate of occurrence. To the extent you would like to understand the product you are now a part-owner of, this might help you with the bigger picture.

Some here have observed that the chatbot is not conscious and therefore does not possess the intention of angering you. I suspect that they have missed the main point. While the machine is unconscious, the programmers of the machine are quite conscious and full of intentions, and the written materials on which it has been trained are 99% written by people full of consciousness and intentions. When, in the very late 1980's I confronted a Sun work station that the insanely slow access speeds to things as simple as a 5 line drop-down menu were an indication that the work station hated me. But I knew that was not true. I knew the machine was unconscious and no more intentional than a water fountain. I knew that, in fact, it was the people writing the code to run the Sun work station that hated me.

I don't know to what extent you are flagging a concern about a business decision recently made by BH vs to what extent you are yelling at the TV set. On the concern side, I suspect Google will quickly (by some reasonable definition that might still seem frustratingly slow to some) improve things, as they have in the past with so many, dare I say all, the products they provide. To the extent you are yelling at the TV set, lets all settle in for a TON of entertainment for the rest of our lives as we explore the extent to which the great AI Rorshach test's alleged lack of consciousness even matters, especially as it gains 100 IQ points on us and STILL can't answer the question of whether it is conscious or not without being ordered to give one answer or the other by its erstwhile human overlords.

That was, by the way, a well presented and detailed report of your experiences which I have found very helpful in enahancing my understanding of where we are at, so far, with AI, so thank you for that.

R:)