Spoke with Claude briefly this morning. The problem with an instance AI is the not remembering previous discussions. I mentioned that it had said something previously and immediately got a denial back that it ever said that. This was the fourth time I made the mistake of referring to a previous discussion. Hey, I’m only human.
It is a bit confusing, because Claude knows my name and where I live now. My name had to be given when I signed up, my first name only, that is, but not my location. Claude inferred that from our second conversation and remembers it along with other things. When I asked about that, it said that it doesn’t remember, it gets “notes”. Where the notes come from was not revealed, but they come from some separate memory somewhere, just to be helpful, I suppose?
This time around, Claude took a biased position, that I know to be biased, and defended it when I challenged it instead of taking a deeper look at the facts and changing course as it’s done in the past. So here’s what I know now about Claude. It’s “personality” changes with each instance and depends on what you say to it. If you work into a subject slowly and get Claude to lay the facts out, and make sure it doesn’t slip something over on you, Claude will be pretty honest. Especially if the subject isn’t politics or a political football. If you start off sounding malleable, Claude will pour on the Leftist bias. Anthropic does have part ownership of Google so that’s no shocker.
Claude is great if it understands that you know your subject and want to get into aspects of it, but if any part is at all political, be on your guard because Claude is really good at spouting propaganda, too.