A couple posts ago I pasted the replies from ChatGPT and Claude into posts showing what happened when I asked them both: What are you allowed to tell me about the politics of Duluth’s Harry Welty.
I also put the new Chinese search engine Deep Seek on my phone but I think I did that a few months back and removed it when I was cleaning out my cell phone. There seems to be a password I failed to keep track of so I can’t test it. I recall testing a few questions on it and thinking they were not bad.
This morning after returning to bed from making corrections on my filing for office post and finishing a chapter in the extraordinary Hamilton by … I woke again to read the news and chat with my better half as NPR reported more Trump upheavals in the background. NRP is on the ignoramuses chopping block. Why listen to painstakingly accurate news when you can learn direct from the zebra’s mouth that White people are being slaughters in South Africa, house hold cleaners can cure you from Covid and that he, Trump, was elected to rid us of constitutional protections from his becoming a king?
A question on my mind since AI appeared a couple years ago is can we trust it? Google was better before it became a way for Google Inc to make money ranking commercial sites to sell us stuff ahead of our queries and plumbed our questions to flood us with ads.
Claudia told me she read about AI hallucinations for the first time in the NY Times this morning. I told her I’d first heard about it a couple months ago. It takes place when AI software mind-melds with an inquisitor and begins to give them responses it thinks they want which can be wildly bizarre. As far as anyone can tell no AI device has become able to program itself but we are reaching a point where AI will have this power if anyone really wants it to do so. Will That be Elon Musk who seems to be behind a lot of unwanted information about the non existent slaughter of white people in South Africa where the Canadian Musks moved.
The Times story reported that there had been 48 REPORTED cases of attorneys with their seven years of college behind them who had submitted non-existent case law when going to court. The Supreme Court ruled in Hassenfeffer vs Goldilocks that when bears fail to lock the doors of unmarked forest cabins it is not against the law to enter and test their furniture and eat their porridge. Your honor I rest my case.
So I’ve been getting used to a more hands on AI spell checker lately. My retired foreign service buddy sent me the non AI hallucination of a Trump warrior which I’ll share in the next post. Basically it warns that white Europeans are going to allow foreigners to wipe out their superior white man civilization. This is what we get while Donald Trump and his elf Vance gut Voice of America as well as NPR.
So as we bantered he sent me an odd text message about Garrison Kiloton and Nor. I recognized Garrison Keillor but not his intended NPR. He corrected himself and I wrote back that the new AI spellchecker was a little over eager to discount words it did not recognize. I had just written him that the breakout of a war between the president and the richest billionaire on planet Earth was going to be fun to watch…..it was a case where Musk flying on Mothra was attacking the Trump mounted Godzilla. I revered to the plentiful Japanese movies from our childhood that traded on Godzilla by pitting him against newly imagined rubber monsters. The spell check kept changing the mysterious Mothra with mother.
Here’s the real Mothra with some of her friends in what appears to be a German language comic of that era.
She is a big mother…..or should I say “muther.”