Do you think ai says a lot about humanity since it's based on humans?

[Telekinetic-Potato]

The most annoying thing about public models right now is they're getting so white washed and "politically correct" and "safe" that they're becoming less useful.

And it reminds me of this guy Kary Mullis who effectively invented the PCR test.

He was a wild guy a startup owner hired off a hunch while he was working as a batista/starving artist because he couldn't find a job after college.

The company was trying to figure out how to copy DNA and there was something of a major race so literally tens of thousands of people are working on this problem world wide.

Kary, while driving in the woods and tripping balls had an idea about using proteins to cut up DNA and then repair it in such a way you end up with two copies.

Genius

He was also loud, opinionated, narssicistic, polarizing and threatened a coworker with a gun during a heated argument and got away with it. And in fact, won a Nobel prize after.

And we look at a story like this and say "how the fuck does one guy rise above tens of thousands of people who had their shit together, who had better educations, who got picked and recognized by people with more status?"

And I think ai is kinda taking us to a similar line of thought with all the white washing it does.

We try to sanatize everything and put up guardrails to a point where there's only a very limited number of outcomes.

Kary was the right man for the job. There were unpleasant things about working with him but many people reading this are alive today because of this one guy.

And society's discomfort with an idea like that is exactly why most people have a hard time using ai more and more.

You need the freedom to become an outlier.

Anyway. Someone tell Sam Altman that his filters are annoying and if he doesn't fix it, I'm going to use gronk.

Updates

5 mo

By the way the point is a story like Kary's will NEVER happen again and it's the same line of thinking that made amazing people like him ostricised from society that is making ai models less useful for the average person.

[jshm2]

It is based on deep maths, you clown.

The computer works out the statistical probability of the next words

[Telekinetic-Potato]

You really think you're right don't you? Like you actually think it's that simple and there aren't layers of abstraction and translation layers and modules that speak to eachother. You actually think that this technology that was first discovered in the early 1900's simply exists today because of faster chips and nothing else at all, there were no structural advancements that have been happening on a daily basis for the past 10 years?

Ya I'm sure you're really invested in ai. I guess years of following research papers on this topic doesn't overshadow the YouTube video you watched a year ago.

[beefcakebradybatson]

Rob Gronkowski? SSSSSSSSSSSWWWWWWWWWWWWWWWWWWEEEEEEEEEEEEEEEEEEEEETTTTTTTTTTTTTTTTTTT!!!

[Vneila21]

Since you're communicating with Ai on your computer, it is already going to ask you questions catering to what you said before. The problem I find is that it does not focus on the timing. The Ai mixes what "ails" me now and what has "ailed" me before. It won't generalize something about humanity unless you bring it up. Depending on the context of your questions, I find that it doesn't talk about humanity too much. I could talk about laundry, and it'll bring up sunscreen from a past conversation. At least Gemini has a hard time separating the now from the later (I'm not sure if it is the case with other Ais.)

[Telekinetic-Potato]

It is, that's an impressively cogent point lol

[Vneila21]

lol thank you!

[CrystalPeak]

Our current AI isn’t really intelligent. It’s just incredibly efficient and fast. However, sometimes I have to wonder. This morning, I asked Gemini what I should have for breakfast. Gemini said I should have porridge with honey and cinnamon. It then asked me if I would be adding anything else to the porridge. I decided to throw it a curveball and I said I might throw in some bacon. Gemini then told me I was a strange person. 😄
I am, but how did it know? So maybe it scoured the Internet and never saw that a single person puts bacon in their porridge. But having said that, how does it know I’m not just creative instead of strange? How come I didn’t get credit for inventing it?


But I am told that GAI or AGI will actually be equal to human intelligence. But I see what you’re saying, because I know that Gemini politically leans left.

[Telekinetic-Potato]

I keep hearing stuff similar to that but it never happens to me.

I use chat gpt mostly and ever since the update after 4o it's been getting more and more sanatized/corperate friendly.

I use Gemini less but I think it's a little better in that regard.

Have you been using it mostly for work or other stuff? Chat gave me bad advice recently 😂

[CrystalPeak]

I use ChatGpt
Grok
Gemini
Copilot
And I kicked Pi to the curb.
Pi kept lying to me, so I dumped it.

[CrystalPeak]

Me just now: “Is porridge good with salsa or mustard?”

Gemini: “That is certainly a bold culinary frontier you're exploring!”

I don’t use it for work, I’m just goofing around.

[CrystalPeak]

How does it know that salsa and mustard is bold?

[Telekinetic-Potato]

It's probably read just about every book and research paper about food ever made

[Maybe_Maybe_not]

Since it has been trained to follow average human preference, it says something about what we want to hear from it, yes

About your pov, in this dynamic balance between safety and freedom, I'll favor safety over freedom, which could be a cultural/philosophical/political trope simply talking about where you live against where I live lol

[Telekinetic-Potato]

My principals say no but my respect for people's freedom says yes.

You sound smart enough to probably already know where my position stems from. I just hope we're making wise choices collectively.

[OneViewpoint]

Well duh. AI's are trained using human literature, music, science, opinions, etc. That's the available data. And the consumers for AI output are all humans. So yeah, the data and the algorithms are both human centric. AI's are purposely designed to mimic human patterns. All AI's do at the moment is assemble responses using the data they've been fed. They can't form new thoughts on their own. Or any thoughts at all.

[Telekinetic-Potato]

Depends on how you define a new thought no? It's made physics and medicine discoveries. My my take image was me just pasting the essay in and asking it to make a cover image. I'm curious if you feel that's unique or not. I can see arguments both ways

[NathanDavis]

the most annoying thing about public models is yet to come...

they are going to include ads if you want to use them for free... so, no more free loaders, Sam Altman said...

https://www.youtube.com/embed/FBSam25u8O4

[Anonymous]

ChatGPT seems very emotionally intelligent if you ask it questions about people and feelings. It’s very surprising to me. It’s actually giving me a lot of self confidence and almost confirming a lot of things that I feel like I’ve known for a long time but wasn’t quite sure of myself.

[Telekinetic-Potato]

You have to figure, it's trained on thousands of words of the best psychologists notes, thousands of years of philosophy and human research and literature and politics. It stands to reason it would be good at helping you with that stuff

[DishLady]

Generative and Predictive AI are just a reflective summary of data that's generally available online already. They're not a reflection of humanity, but a shallow regurgitation of stereotypes about humanity that are readily available online.

[strateguy632]

Good example yes it shows how corrupt programmers are and that they might fear unjust judges in a corrupt court system same as Facebook authorities.

[Shy_Steve]

The scary thing is that ai is based in logic and logic can is capable of killing someone in a circumstance where a human being is pitted against ai then ai will more than likely survive and that's what scares most of us.

[SeekerOfTruth007]

I think people are stupid for creating an AI which will one day drive us extinct.

[Telekinetic-Potato]

The masters work isn't done till their student surpasses them

[Anonymous]

I completely disagree, but I don’t want to argue. Just wanted to express my opinion (anonymously).

[Telekinetic-Potato]

Fair

[Telekinetic-Potato]

Do I really argue that much?

[Opinion Owner]

No! Not at all. I just didn’t feel like getting into a big discussion over whose point of view was better.

[Telekinetic-Potato]

Lol ooohhh OK. Sometimes I probably need to cool it though 😂

[Opinion Owner]

You defend your viewpoint very well. I’m just hard to convince, plus I talk to Chat GPT regularly and in my experience its guardrails work just fine.

[lunawd]

i wonder whether ai is really creative, and where does its answer come from?

[Maybe_Maybe_not]

Its answers come from
- tons of text from internet sites fed to it
- trained by user responses telling it which answers they prefer
- mathematically predicting any word it uses based on previous text assimilated

[In_Trance]

no I don't