I’d suggest anyone wondering about this to look up the ELIZA effect, it’s interesting and outright terrifying at the same time, and more relevant than ever
5 Likes
Careful. It’s someone’s waifu you’re disparaging here.
4 Likes
meh it’s for simplification, i don’t understand how says that AI thinks or is trained makes people make it more human, saying “generates off weighted data” is just too long, probably english don’t have alternative ig…
I’d say speaking of a machine that isn’t sentient or able to think as “thinking” would make people think of them differently and make people less likely to understand that they do, in fact, not think
The ELIZA effect is a perfect example of how this is a problem and the issues it can cause
3 Likes
i think it’s only problem in english, as thinking is too general for both sentient thinking, processing information ect.
also the problem is that AI gives nice illusion of being too real, while in reality it uses algorithm you could write in 1950s, while scaled it gave very impressive results + came out relatively quickly, so people are hyped about it being the future, smth like flying cars.
Not really it’s an issue in other languages as well
No the problem is partially that people view it as sentient, as demonstrated by the ELIZA effect where there was no real level of sophistication involved, yet people still anthropomorphized it because it seemed human in the way it responded
The level of seeming quality of the output isn’t really the thing that makes people think it is able to think, it’s that it “talks” like a human
The way to deal with that is to guard oneself against thinking of it as “thinking” to not fall into that trap
4 Likes
meh, language trap, real problem could be fixed if people were taught how programming works and how AI works under the hood, because then it would make sense to them that if we scale it up by milions it will talk like human, when in reality it will be just emergent behavior of how it works on micro level
also what do you expect? even most known scientists have problem of defining consiousness, cuz it makes no sense if we think about it, normal users don’t have study in philosophy and AI engineering usually
Sure but these LLMs are not sentient, or conscious, so that’s not really relevant
That’s not realistic, and wouldn’t be needed if people stopped talking about it as if it is thinking
Also that knowledge isn’t actually sufficient, you need to consciously view it as non-sentient because of the ELIZA effect and general human nature, plenty of people well aware of how LLMs work fall for the trap of thinking they are sentient
4 Likes
for now - ig no one knows, time will show, thx for talk tho
Can the technology achieve sentience in the future? Who knows, I doubt it and so do a lot of experts
But the current technology isn’t, plain and simple
Cheers!
3 Likes
Sentience is one of those words with no clear universally accepted definition. Like species, gender, free will, culture or self-awareness. Remember the Turing test? AI passes that now but it is now considered that the test was flawed. For example, is a worm sentient? Or a bacteria?
Not really, it cannot appear entirely like a human under all testing conditions, and fails the Turing test
It’s not just “seems human”, it’s “indistinguishable from a human”
1 Like
Hmm. There seems to be some disagreement on that. I binged it and yes the AI answer said no. But here it says yes.
Not that I am that bothered about it really. I think the test is flawed IMHO.
For most people, language constructs can form their reality. That’s why calling it “intelligence” paired with the “proof” in the form of human-like text output, conjures a (false) reality that it’s an autonomous intelligent entity behind it. Weaving realities from words is known as “word magic”. It can and often is used with malintent. In fact it’s a mass manipulation technique as old as human civilization.
7 Likes
The Turing test was invented by Alan Turing in 1949. It’s no longer held as a valid test of human level intelligence. It has remained popular because of Blade Runner. Basically, it’s just not complex enough.
2 Likes
The Turing test is just a test where the machine passes if the person it interacted with is convinced they are dealing with a human. It is a milestone at least, and has already been passed by current generation AI, and so, yes its a bad example of how to measure machine intelligence. Same with Wittgenstein, as certain statements can be twisted to seem like language and soul are connected.. The general trend is to conclude that a machine can not be intelligent …
Because most people using that type of language don’t have the knowledge to differentiate between metaphorical statement vs really anthropomorphizing LLMs and other so called AI products.
Cheers!
2 Likes
Clearly I have stayed off of this topic for a while.
Training, as a variable for “LLM providers/admins setting different parameters and giving the LLM different ‘references’ for the ‘best’ outputs”,
being overly descriptive on things i do not care about is not my style, even if it is supposedly “anthropomorphizing” it.
I am not going to waste time on using the “correct” terminology for the rules being set for some “so-called artificial intelligence” merely because everybody else cares about people’s viewpoints, or more specifically the viewpoints of those people who are clearly inadequately educated on glorified python scripts.
I am truly confounded at the meaning of this, but I find it somewhat amusing.
Do tell if you will.
key word, Most people. Language is not reality, it is a way to better preform actions in reality.
For me, it means nothing other than : I am tired, lazy, and/or I simply do not care.
At least one person came to that conclusion.
Again, most people do. I do not, nor do I care in the slightest.
What also, does it even solve for people knowing their “imaginary friends” arent sentient?
From what I see, absolutely nothing.
Most actions made by humans boil down to a maximum of 2 reasons; emotion, rarely logic, or both.
Feel free to ignore this post; I know many will either way.
I’m an amateur game developers, it will spend too much time to turning my imagination into real game if no LLM. But before vibe coding, I will discuss a suitable plan with LLM about how to construct scripts. So my views is that discussion on the forum could focus on ideas and construction of different modules. For debug of detailed scripts, LLM maybe a more ideal tool.
What’s wrong with taking the time to learn to do something well?
3 Likes