Human

(quarter--mile.com)

Show context

notepad0x90 ◴[15 May 25 04:54 UTC] No.43991964[source]▶

Why are emotions so special? they're just algorithms like any other. Emotions aren't what make humans different than machines. feeling something is similar to an LLM model reacting to a prompt a certain way. Just because chatgpt is trained to not "feel" anything (to avoid controversial output) doesn't mean LLMs can't feel things like we do. self-awareness, self-training, adaptability, original thinking, critical thinking,etc.. are different questions. but I see no reason why machines can't receive input/stimuli and react/output by the same way we do because of how they feel about the input.

replies(5): >>43992068 #>>43992108 #>>43992143 #>>43994759 #>>43995050 #

1. jplusequalt ◴[15 May 25 13:19 UTC] No.43994759[source]▶

>>43991964 #

>Why are emotions so special? they're just algorithms like any other

Nobody understands what emotions are. Nor can they predict which emotion someone will feel in a given situation, or how they'll act under the influence of that emotion. Emotions aren't the mechanism by which humans solve problems, and rather they are often an obstacle to overcome. Emotions also aren't "finite" or "rigorous" as those terms aren't applicable to ephemeral phenomenon.

This is the kind of confidently incorrect statements people who work on software say that irks me. Not everything in life has a nice and simple parallel to computer science. Just because a person can abstract about one subject well, doesn't mean their tools of abstraction can be applied to all other subjects.

replies(2): >>43995199 #>>43995273 #

2. notepad0x90 ◴[15 May 25 14:01 UTC] No.43995199[source]▶

>>43994759 (TP) #

I didn't claim to understand what emotions are but that by observation we know they are deterministic and a result of belief and memory. they're not magic. Just like how I can't tell you what the specific configuration of chatgpt's model is but I do know it is a model that is made up of memory (public internet data scanning) that is trained by prompt engineers (belief parameters). Emotions may be dynamic, unreasonable and similar terms, but they are finite and rigorous, it is just that we struggle to fully grasp them due to the immense and complex nature of the human mind.

> This is the kind of confidently incorrect statements people who work on software say that irks me. Not everything in life has a nice and simple parallel to computer science. Just because a person can abstract about one subject well, doesn't mean their tools of abstraction can be applied to all other subjects.

If I was a sailor, I'm sure I would be using sailor metaphors and analogies. the message of my comment and the facts of the matter don't change either way, whether it irks you or not.

replies(1): >>43995466 #

3. bsza ◴[15 May 25 14:09 UTC] No.43995273[source]▶

>>43994759 (TP) #

No one can predict what an LLM will say in a given situation either, except by running it. No one can even predict what a double pendulum will do next. If anything makes emotions exclusive to us at all, it’s certainly not predictability.

replies(1): >>43995437 #

4. jplusequalt ◴[15 May 25 14:24 UTC] No.43995437[source]▶

>>43995273 #

Okay, semantics aside, my comment was getting at something else. I'm arguing against a certain kind of reductive viewpoint software peeps tend to indulge in.

replies(1): >>43995633 #

5. jplusequalt ◴[15 May 25 14:27 UTC] No.43995466[source]▶

>>43995199 #

>but they are finite and rigorous

Bold claim, yet you fail to demonstrate this.

>the facts of the matter don't change either way

What are the facts? It seems to me you're just spit-balling.

6. bsza ◴[15 May 25 14:46 UTC] No.43995633{3}[source]▶

>>43995437 #

I agree with that sentiment, I just don’t see OP making that kind of reductionist claim. They said emotions are algorithms, not that the algorithm is simple or even deterministic.

↑