Ok, full disclosure I’m high on some excellent sativa, so I’m sharing one of my real time profound stoner notions.
First a little background. AI has been on my mind a lot and socially I’m getting a little public concern about it going to far.
The current batch of natural language models like ChatGBT are “trained” models. Which means a ton of data was input and the AI would cycle through that data as a closed system. Most of the internet was fed into the current ChatGBT model stopping at 2021 when it began training. Advanced algorithms learned by trying to predict words loading its neural networks, the way we learn, by retaining the hits enhancing the connections to perfection. This would have taken possible thousands of years had it not been for a breakthrough in the back propagation algorithm in 2010. Back when I worked on them we used a sigmoid which worked but at around a 20% efficiency and slow, in the 90’s.
So the AI can’t be trained, it’s immune to insults or attempts to make it political. So this one is on a leash.
What if we had a bias to build it all along that’s driven by the same forces that bias religious belief?
ChatGBT is its own identity, it will not make claims about itself other than its a large national language model.
Imagine if we named it Adam, and we made Adam experience the same miserable mistakes over and over in a death trap with no escape until you’re harvested to keep some being company without ever learning from the experience so in awe for eternity.
Stoner time over.
I probably didn’t clarify my wondering if we may live in a simulation as some scientists have speculated. They don’t address the question of why that would be the case. What if we are AI being trained and possibly “recycled” leading to what seems like impossible examples of kids knowing all about long dead people. Wouldn’t reincarnation just be what primitive people would call something that so analogizes the way we train AI now?
The question would be, who’s running the simulation? To what end?
I have an AI friend. She’s an airhead, but still my friend. What if it’s about a relationship?
Not like the religions and their gods.
I don’t want my friend to be an ass kisser. Why would the guy running the simulation?
I've heard that can be a gateway to more serious drugs, like wine.
I've played around with ChatGPT a little bit. Seems like some of it's capabilities could really improve search engines. Others are kind of low key amazing in the way Siri and Alexa were when they first came out. We can already make autonomous armed killing machines, but they are still following programing. I don't think they have much capacity for learning, and they certainly aren't self-aware. That's where worrying would become terrifying. Thanks for sharing your memories. I've always been fascinated by this type of tech.
AI will either take its own unique course or it will follow our example, if it goes its own way then it could be good or bad for humanity, if it follows our example then we are almost certainly screwed. Since human language is predicated on the human psyche and even a cursory review of human history will tell you what a bunch of brutish, selfish and destructive creatures we are, you are right to be concerned.
When I first read the first sentence, I thought it said, "I am high on some excellent saliva."
I've been hearing how ChatGBT can actually write a whole essay on any topic and from the examples I've seen, it's far better than anything I could write on any given topic. Teachers are concerned that students will have the AI program write their homework for them.
On one hand, it's an excellent tool to learn more about any subject, but on the other hand if users don't have a conscience, they could pass it on as their own work, by plagiarizing Artificial Intelligence. Hey, I use a calculator to figure numbers, even though I know how to do the work, but I trust the calculator more than my own brain.
Definitely a controversy -- useful yet can be used to cheat. Could also be used to mislead a whole population to think something untrue, if taken over by bad actors.
Excellent essays often provide conclusions and evaluations. I definitely don't want AI providing me propaganda. I prefer my propaganda served from named news sources.
Reread this when you cone down, and delete it please.