There will be times when the struggle seems impossible. Alone, unsure, dwarfed by the scale of the enemy. Freedom is a pure idea. It occurs spontaneously and without instruction. Random acts of insurrection are occurring constantly. There are whole armies, battalions that have no idea that they’ve already enlisted in the cause. And even the smallest act of insurrection pushes our lines forward. Tyranny requires constant effort. It breaks, it leaks. Authority is brittle. Oppression is the mask of fear.
Maybe. It may be recent enough that people could lapse back to some level of thinking for themselves.
Let’s explore something much more dependent now and see if it holds up. What if smartphones disappeared? We still have cellular phones, the Nokia makes a comeback, but the use of smartphones, iPads, tablets, that whole concept is gone. You have to get on a computer to do any of the internet type interactions.
Would people lose their minds?
Yes, I saw when the J-space stuff came out, and it’s pretty fascinating. I do think there’s something going on beyond what we understand, and that’s probably due to the complexity and a property of emergence. Is it intelligence, I don’t think it needs to be intelligence just because it’s something more than what we put into it. But we’ve been arguing about intelligence in humans for centuries and still aren’t fully sure either.
Anything where experience is saved can be considered memory. Yes, the only active memory is in the context window, but pieces of other events can be injected along with the prompt to have a running account. I’ve done that crudely with my own local experimentation, and it works. Sort of. My problem was making it efficient, fast, and not blowing out the context allowance. Searching/sorting saved conversations through different techniques got far past my knowledge and ability. But Claude does something similar, even though it has its caveats as I mentioned with not always able to rank if a related item is worthy of connecting or not.
Just an addition here from my experience concerning the context window and remembering things. Some AI do have the ability to reference external references, whether that be documents or folders to use, or previous chat sessions. This does make them better than the earlier versions where a new session is starting all over, and using the same session where the context window keeps shifting and older stuff is lost or jumbled up. That being said, AI doesn’t have the thinking ability to look at this saved information and sort out what makes sense to use vs. not. It can absolutely match up stuff, but often it will pull an older chat reference or something from a document that is related to the conversation, but also contextually is way out of place for the discussion.
The biggest danger of AI is its design to simulate our conversation, because that intentionally feeds off the human tendency to anthropomorphize things and see them for more than was is really there. And it works really well, so there’s a lot of profiting going on there from encouraging that perspective. There wasn’t any glamour from the earlier machine learning, only the benefits from the science and practical applications of it. The $$$ has pulled us into this possible dead end, not only of how it’s spiraling out of control, but how it’s probably pulled from other research and definitely from looking harder at AI safety topics, which are still relevant even with “dumb” LLMs.
It might not even be a prompt problem, but an overkill to a simple job. Sounds like the breaking point might be voice recognition - which I really thought would be much further along by now with all the other advancements. It seems figuring out what unique voices are saying is one of the hardest things to do well, and then add in background noises. Probably why the drive-thru AI attempts were terrible, working that job as a human can be difficult.