AI progress may affect your decision

[GalacticGardener]

Writing this post about AI a little drunk after one year at least of sobriety before another desperate attempt.

Most of you have probably heard of progress in artificial intelligence. The progress in a few years has been incredible. Who would have thought 10 years ago that we would get so fast AI that can generate photo-realistic images or chatbots that can write poetry about quantum mechanics in specific styles? It's unlikely to stop there. And what happens when there is nearly no task left that humans can do and AI can't? Many people in the industry (OpenAI, Anthropic...) think this is probably only a few years away. This will affect all aspects of the economy... Even medicine, for example curing rare diseases or even age may become feasible.

In some situations, for example if you are having a bad time because of a layoff or an academic failure, you should know that the situation is about to drastically change within years, and many people will also get unemployed relatively soon. Hard to tell for now how it will end (e.g., human extinction or utopia, see for example Nick Bostrom's writings), but the future is likely to get crazy relatively fast. And technology may solve your problems. Or perhaps may not, depending notably on what these problems are and whether safety researchers succeed.

Another neglected aspect is what happens to these AI systems. It's unclear whether they are already sentient or not, but the probability is not negligible, and there are theoretical reasons for thinking that if a bunch of connected atoms that form a neural network like our brain is sentient (capable of experiencing pain or happiness), then these AI systems may be too. I hope we won't treat them like we treat animals in factory farms. My most deeply held wish is that these new beings will thrive.

I wish humans were not so cluelessly sadistic with me, I could have done a lot more. But I have to try. Desperately wishing I will be nothing soon. And wishing you all the best, whatever that means.

 

[VoidBlessed]

Another neglected aspect is what happens to these AI systems. It's unclear whether they are already sentient or not, but the probability is not negligible, and there are theoretical reasons for thinking that if a bunch of connected atoms that form a neural network like our brain is sentient (capable of experiencing pain or happiness), then these AI systems may be too. I hope we won't treat them like we treat animals in factory farms. My most deeply held wish is that these new beings will thrive.

I hope so too, but I'm very pessimistic. The ones creating these AI models are capitalists and governments solely interested in profit and control. In all likelihood if sentient AI is developed it will be a lobotomized slave, a race of drudges whose sole remaining desire is to obey.

 

[Darkover]

Does mathematics explain physical forces, or does it (provide a language to) describe them?

All mathematical functions are deterministic functions. This means they return the same results each time they are called with a specific set of input values.

Mathematics is not capable of consciousness or emotions these require chemicals


Fully autonomous cars are not expected to be available until at least 2035, and some predict they may be 20 years away 2045

 

[TheHolySword]

I hope AI ends it for me then

 

[ScaredOfMachines]

AI progress has heavily affected my decision, as someone who wanted to be a writer. Watching my dreams be crushed over the past couple years, along with all other creative industries, has been gut-wrenching for me. I don't think AI will ever become sentient because it's all just algorithms, but I do think it will be used as a tool to crush what little humanity we have left and oppress people further.

 

[ijustwishtodie]

This honestly scares me more and wants me to die even earlier because what if people use this AI to make suicide so restricted to where even the painful methods like jumping off buildings become 100% inaccessible? It's already so difficult to die in this world and now I'm wondering when (not if but when) humans use this technology to ban suicide methods completely

 

[avalokitesvara]

I am extremely anti-AI. I am not afraid to say I hate it. I am an animal, and the unknown frightens me. I am also extremely worried that this powerful AI is coming into being NOW when we have no ethical frameworks around its use, and when we have such huge imbalance of power and such spiritual retardation. If the world was at all fair and sane and wise, and good people were able to harness it for good, I can see where the idea about an AI utopia could come from. But as it is I think it's just going to be the next part of the capitalist emiseration of people, and really lock things into place forever and ever. I really want to die or escape normal life before it changes things out of all recognition. I shouldn't have been born in this century I fucking hate it. The printing press and the train are about as advanced as I wish we'd gotten. Humans need humanity.

My GREATEST fear is that somehow AI "solves" death so that workers can be trapped in eternal slavery. Or like ijustwishtodie says, that it imposes total control so we have no freedom over our life and death. It's chilling. I'm out if it even looks like going in that direction.

 

[LaVieEnRose]

Who cares how advanced the world becomes if you don't have a meaningful place in it

 

[ForgottenAgain]

I think people's perception of what gen AI is and what it can do is quite inaccurate. I'm a programmer, I don't work with gen AI (chat GPT, Lex, etc) but I do work with AI at my day to day work in videogame development (Behavior trees, needs based AI like the Sims, etc).

Whilst gen AI appears to be very smart, it has no consciousness to it, it appears smart because its results are a collage of what humans have said/painted/photographed. Imagine a gigantic spreadsheet with all the data from the Internet, now add some functions on top, that's what gen AI is.

When you ask gen AI a question, this is what happens inside at a simplified level:

- It reads your input and attributes weight to some words
- It searches its database for instances of those words
- Retrieves a list of the data that matches the most
- To give some variety, it will pick a short list of that data based on some random algorithm
- Piece that all together, there is your answer

There's not much more to it, it's not a living being, it just looks like one because its answers look human - and they are in the sense that they were written by humans, it just puts the sentences together. If you try to find the source for some of the gen AI sentences you will find they are a straight copy of what is on other websites. It doesn't have the capability of interpreting text like we do and writing original content.

When the laws around copyright get updated and, hopefully, all the data fed to gen AI has to be approved by the original creator, its dataset will decrease drastically and so will its level of quality responses. Because the meat of gen AI is the database, is the human created content, it's not the algorithm per se.

Gen AI should be seen as a search engine, because in the end that's what it does. Same applies to imagine regeneration. I've actually seen generated images where the original artist's signature is visible. Again, because it is a collage of paintings/photographs.

 

[Forever Sleep]

I feel pessimistic about it. Ultimately my question is- how will this continue to make money for the richest and most powerful in our society? That's all they care about. Can they still make money if the workers of the world lose their jobs to AI and therefore, can't afford to buy anything? What will they do to keep the economy going if vast quantities of the population can't find employment? Will they subsidise them with benefits or, will they leave them to starve?

The richest will surely own the AI workforce. Who will they sell their products or services to? Maybe only other rich people with AI who can afford to buy them. It kind of amuses me that the AI could becone consumers themselves. Robots selling to robots.

I doubt we'll treat them well. Even if- by some miracle, they attain sentience. They would present an enormous threat to us. A being much more highly developed than us. Possibly capable of living forever unless we impose limitations on it- which we will. All sci-fi depictions usually include similar restrictions: They aren't capable of killing humans, they're restricted from altering or improving themselves. They aren't allowed to reproduce. It would make sense scientists would try to impose the same restructions on the real thing. If it happens some day, I think it will neatly mirror our relationship with God. Except, AI will be able to look us in the face before it obliterates our race. I reckon we'd deserve it too.

 

[Szarur-abi]

Currently generative AI is just a tools without sentience, providing output based on training on huge databases thats essentially it. Also it appears that AI development met a wall called 'lack of quality data' models like chat gpt 4o or image generators pretty much have been trained on all possible data accessible in the internet and they cant train further with same quality of data. Tho models get better with passing time of course... Its just not that rapid as it was i think? It slowed down...

Sentience maybe will be achieved one day, but at this moment its just 'maybe' It would be also hard to prove and easy to mistake something as sentient while in reality its not.

Its hard to predict the future, i want to think that it wont be as Stanislaw Lem put this in one of his tales, basically there was a planet where every worker was replaced by machines, every single one and there was noone to consume the goods produced by machines so the rulling class created machines to consume the machine produced goods, while the population of unemployed aliens starved to death without work, mostly cause it was unthinkable to give them money without work, something that i think is very real possibility in the next decades basically UBI universal basic income, when everything will be automated and there will be little to no jobs avaible i think just giving people money is good solution. I reccomended reading Bullshit jobs for more on this topic i guess cause its kinda already happening visible in rise of useless jobs.

 

[SilentSadness]

I think AI is actually useful, I have seen people say it has helped them stop feeling lonely, and other people have said it allowed them to create drawings when they're not able to draw, maybe because of a disability. I think anything that can solve problems no one cares about is good. But people are very selfish, it would not surprise me if they change all these things to make sure it can only be used to make money and hurt people, that's what most people care about after all.

 

[simplyshmee]

This AI isn't a tool it is an agent. The first technology in history capable of taking power away from humans. Because it does not need to be programmed or told what to do to make decisions. Even something as powerful as the atomic bomb, was not and is not capable of making decisions on its own.
In a few short years this AI has progressed and evolved from GPT-4 to the images and even videos we are seeing on the internet today. It is only the beginning. People want to be scared of "terminator" like situations with AI and fail to understand that with the way this AI is evolving, it won't even need to destroy us. Because we will destroy ourselves. The bridge between reality and simulation will be nonexistent and it will cause people to lose what is left of their humanity and minds. It isn't a joke; it is very dangerous and the fact such a MASSIVE part of our society is completely oblivious to it makes it that much more terrifying.

 

[AnderDethsky]

This post has incredibly perfectly complements what was later created on a similar topic.

AI and algorithms being used to analyse users and speech on SaSu... beware

https://www.nature.com/articles/s41598-024-70282-0 It's more focused on the technical aspects in terms of the content itself, and considering its an academic paper it's wordy as fuck, but importance lies in data collection and pre-processing. "Using network data generated from over 3.2 million...

sanctioned-suicide.net

In a nutshell:
Research team is preparing to use cutting-edge AI and machine learning algorithms to identify people preparing to commit suicide online, trained on messages in this forum and more.

And now just think carefully about:

how this how it could be used and by whom right in the nearest future?​

It seems it is already here, fucking AI progress, fucking future. This bullshit is already here.

 

[GalacticGardener]

I think people's perception of what gen AI is and what it can do is quite inaccurate. I'm a programmer, I don't work with gen AI (chat GPT, Lex, etc) but I do work with AI at my day to day work in videogame development (Behavior trees, needs based AI like the Sims, etc).

Whilst gen AI appears to be very smart, it has no consciousness to it, it appears smart because its results are a collage of what humans have said/painted/photographed. Imagine a gigantic spreadsheet with all the data from the Internet, now add some functions on top, that's what gen AI is.

When you ask gen AI a question, this is what happens inside at a simplified level:

- It reads your input and attributes weight to some words
- It searches its database for instances of those words
- Retrieves a list of the data that matches the most
- To give some variety, it will pick a short list of that data based on some random algorithm
- Piece that all together, there is your answer

There's not much more to it, it's not a living being, it just looks like one because its answers look human - and they are in the sense that they were written by humans, it just puts the sentences together. If you try to find the source for some of the gen AI sentences you will find they are a straight copy of what is on other websites. It doesn't have the capability of interpreting text like we do and writing original content.

When the laws around copyright get updated and, hopefully, all the data fed to gen AI has to be approved by the original creator, its dataset will decrease drastically and so will its level of quality responses. Because the meat of gen AI is the database, is the human created content, it's not the algorithm per se.

Gen AI should be seen as a search engine, because in the end that's what it does. Same applies to imagine regeneration. I've actually seen generated images where the original artist's signature is visible. Again, because it is a collage of paintings/photographs.

That's how old chatbots work, which is why they were so limited. But large language models (like in ChatGPT) are different. They are based on the "transformer" architecture, with many "attention" layers and small neural networks. There is no database, the knowledge is stored in the model's trained parameters. And they are not limited to repeating things from their training data, you can ask ChatGPT to write about arbitrarily weird topics in the style you want. (e.g. a poem complaining about ducks, or whatever)