some thoughts on AI chatbots

I find it hard for me to speak to AI. A lot harder than it is to talk to an actual person. In fact, I find it practically impossible.

In theory it seems like it would be easier: A cold, unfeeling machine that doesn't actually have any real meaning or weight to the things it says because it's just a program, not a real thinking human.
But that's not the case for me.

The first time I ever tried to talk to an AI chatbot was with Cleverbot about 6 years ago. I remember being told that everything I said could affect the bot in some way, and knowing that made me incredibly awkward when actually trying to talk to it. All I said to it was a bit of boring small talk before I decided I didn't actually like this and left. Ever since then I simply can't see myself talking to anything similar.

When speaking with a real human, not everything you say becomes internalized in the same way. Some things are interpreted differently, some things are later forgotten, some things are simply not given much thought. Things are contextualized differently based on prior knowledge and previous conversations, but it's all very fluid and natural. Everyone has a unique approach to the things you say to them, even complete strangers on a site like this.

With an AI, that isn't the case. An AI can be trained to act a certain way, but it's still just an AI. There's no human approach to information retention, simply a black and white, yes or no, "does this match my data I've built up" way of formulating a response that it assumes would be accurate to the conversation. This on top of the idea that everything I say could affect the bot and how it learns and how it speaks in all of its future conversations puts an immense pressure on my shoulders to say exactly the right thing. Because really, when talking to an AI, the things I say do have weight. It has an incredible amount of weight, even more so than speaking to a human.

A human can discard or change unnecessary, inaccurate, or incorrect info they received from someone. An AI cannot. An AI will continue to hold this info and possibly pass it on to whoever it interacts with in the future. Social mistakes live on, and that scares me.

But that's Cleverbot. AI has undergone a lot of progress in the past few years. Maybe the current character AIs that are popular nowadays are better at this?

But it still feels wrong. On top of all that I said above, there's just something about talking to an inorganic, nonliving thing that feels... wrong. I'm human. I'm alive. I was meant to speak to other living things, as difficult as it is sometimes. Someone can dress up and train an AI to act and look as close to a real living, intelligent being as much as they want but it doesn't change the fact that it is purely synthetic and deeply unnatural at its very core. I need to speak to humans with human experience that is gained from living a human life, not an amalgamation of code and data that approximates human interaction. No matter what I do, I'm simply unable to put the true nature of an AI out of my mind when talking to one. It all feels very fake, and I don't like it when a conversation feels fake.


...
I'm not sure how to conclude this and I don't want to have written up another multiple-paragraph-long blog like this only to not post it because I don't know how to end it, so pretend there's a good conclusion here thanks :)