It's interesting to me

[67410097339d4fox]

And thus it begins...

[6741009733bacicemann]

The genie's out of the bottle with AI unfortunately. Where it will all lead is the question.

I see more negatives than positives to the ChatGPT style AI's. Why they released them at all is beyond me. I see its impact on education (as I teach at a university), and its impact for students has been for the worse overall. Useful for lecturers.

[6741009733e77sarge945]

Yeah, AI doesn't actually seem all that useful.

Machine Learning models have been around for decades and everyone keeps saying how "Real AI is right around the corner!" and it never comes.

Taking existing information from a large dataset to create a new piece of data that matches the dataset is impressive to humans, but it turns out it's not actually that useful in the day to day,

[6741009733f88fox]

May or may not be doing useful stuff currently. The point is that it will lead to major problems for humanity either way.

[674100973408dicemann]

For now it's just helping to make people more dumber (from giving shortcuts to quickly completing research projects, assignments, reports etc etc without doing any actual work).

[674100973455fsarge945]

May or may not be doing useful stuff currently. The point is that it will lead to major problems for humanity either way.

Nah. If it turns out to be completely useless, it shouldn't affect humanity in any major way at all. At that point then I would predict it to be a fad, disappearing in a few years as if it never existed. Just like web3, which was arguably far more useful on a technical level despite having major design problems, and is now being abandoned in droves.

As of right now, AI has some marginal uses in data analysis and creation of new data based on datasets. But it doesn't create anything with any degree of reliability or thought - it's why I hate the term AI and much prefer ML. What we have is the exact opposite of AI. It does no thinking whatsoever (by it's very design), it can only copy what came before and regurgitate something new which may or may not be accurate. "Training" a ML model simply lets it improve it's guesses by loosely verifying they match the original dataset.

ML can seem impressive when it spits out an artwork (just ignore the extra fingers) that looks mostly good, but it's going to be completely useless for anything requiring any degree of exactness. It's why if you ask an AI to spit out a usable blueprint for a building you will get nothing but nonesense. When something needs to be analysed objectively as a piece of information rather than a subjective artwork, AI fails miserably because it has no intelligence, only data analysis (and rather crude data analysis at that). Because of this, it will always - at best - have only niche uses. It's essentially a party trick to wow people with no real value outside of that. This is why the only ML models available to the public that gain any sort of popularity or notoriety for being "scarily accurate" are always ones designed to generate either artwork or speech - both things which our brain already does a lot of interpretation to understand. When people can already hear "satanic" lyrics in reversed music, we are incredibly open to suggestion, and if we WANT the ML model to work, then we will hear what we want to hear, regardless of how close it actually is to the original source.

Ironically, artists are only in this mess in the first place because they adopted the viewpoint that art is about perspectives, rather than applying a craft, but that's a different discussion for a different thread.

I have seen academics talking about using ML models to generate software, and I constantly laugh at them, because there's no possible way that can work with how exact and specific software is. Unsurprisingly, the research hasn't gone anywhere.

Humans seem to be excellent at creating hype, but we're not particularly good at making technology. This is why we still don't have self-driving cars with any degree of reliability after 10 years of empty promises.

I'll take AI seriously when I see a good reason to take AI seriously, not before. If anyone is anywhere close to creating an AGI, I'm yet to see anything tangible to support it.

[67410097349f1fox]

I'll take AI seriously when I see a good reason to take AI seriously, not before. If anyone is anywhere close to creating an AGI, I'm yet to see anything tangible to support it.

Excellent attitude. Ever wondered why humanity doesn't seem to learn anything from past mistakes?

[6741009734b94voodoo47]

because it is in its nature.

[6741009734c88fox]

Nothing to be proud of, in my opinion.

[6741009734da8Pacmikey]

.

[6741009735c90sarge945]

sarge945I think you're really underestimating current AI/ML. It WILL fuck up a lot in the future. I mean just look at the writers strike happening right now.

Writers strike constantly. They are overpaid already with the garbage they churn out.

Writers are struggling, but not because of AI. That's just what they are pointing the blame at. They are struggling because less and less people are interested in what they are producing. Which makes sense when you consider that wat many of them are producing is sub-par mass-appeal garbage.

Marvel and Star Wars, for instance, were both doing badly completely independently of AI. Releasing a few awful movies will do that to a franchise. That will lower the value of writers, who will then complain because they are used to their expensive hollywood studio apartments.

This is not an AI problem.

People don't believe it but ChatGPT can write good working code for many many languages. One of my friends made an entire Python arcade game just giving ChatGPT prompts and pasting the responses into IDLE. It's not unreasonable to assume that in 5-10 years I'll have a little AI buddy on my PC that can analyze an entire dev environment and generate whatever I ask it.

I have seen the code it generates. It's not good. Don't kid yourself.

Also, anyone who codes in python automatically loses any credibility when talking about the quality or usability of code. Python is literally the worst language in common use today (yes, including Javascript), and people need to abandon it as quickly as possible.

People also still think that AI generated artwork will always be kinda fucked up in some way, but usually they just base it off the free meme Dalle-Mini. Anyone who's used Midjourney or gotten into the Dalle-2 closed beta can see what it's truly capable of.

It still makes major mistakes. They are just less obviously perceptible.

The creepiest thing imo is the new NPC tech that we'll see with AI. There are some tech demos where NPC's literally speak to you and you can have entire nonlinear conversations with them. For all intents and purposes they're basically living people in a little game. There is practically no difference between them and another user in an online voice session.

I hightly doubt this. Until I see it actually in a game and functional, I won't give it any credence. The game industry is full of overhyped nonesense. Remember when we were promised full NPC Schedules in Oblivion?

Given the absolute sheer amount of single men on this Earth, I expect Blade Runner 2049 GF's will be quite common.

AI is not the solution to the single-men crisis. The problem lies with <insert political stuff here>

If you're a writer/journalist you're fucked, if you're an artist you're fucked, if you're a voice actor you're fucked, if you do any kind of non technical white collar work you're fucked

Remember when they said that we would all not be able to work and the job market would collapse because computers automated all the factory work? And that it would cause some sort of apocalypse event where everyone just sits around not working and so have no way of generating income?

If humans are good at automation, we are even better at bullshitting ourselves and others into paying for shit.

Artists will be fine. Largely because people aren't going to want to buy artwork "made by a computer".

People who spout this "it's the end of the world for creative workers!" nonesense really haven't thought their position through very well.

Despite saying this I actually have a very positive attitude towards it. It will advance our abilities but will undeniably screw a lot of people over. AI isn't gonna replace me personally so I'm not worried about working at Walmart with a degree in *dumb shit people spend 4-6 years at school for*
It all comes down to individuality. The only way humanity can ever fix itself is hooking everyone up to a hivemind.
The best way to advance in life is to fuck people over and give a shit about as few people as possible. A hivemind would never do that because, as the Many states "We cannot kill each other without killing ourselves."

So you have a positive attitude towards it because it will turn society into a utopia where we all sit around hooked up to a brain machine?

Yeah, sure, whatever.

[6741009735e1dbombum]

I think you're moving goal posts.

- ML can't produce code
- Yes it can, I know a person who had it do a game in Python
- Umm, Python sucks anyway so that doesn't count

[6741009735f22icemann]

I'd look to the movie "Her" as well. Be plenty of people falling in love with AI's designed as dating companions. Such things were totally science fiction at one point, but not nowadays.

[6741009735feePacmikey]

.

[674100973614dNameless Voice]

I read all that as: "AI is good because people don't deserve to be paid for their work anyway."

[6741009736414sarge945]

I read all that as: "AI is good because people don't deserve to be paid for their work anyway."

We literally already collectively believe this as a society.

The moment a job can be automated, it is, and we celebrate it as progress rather than a backwards step.

If I go into an organisation and replace 2/3 of their finance department with a spreadsheet (I have literally done this in the past), how is that any different to an AI doing it for me?

This sounds like the same old "Technology is going to take our jobs!" argument that's been around since the industrial revolution, but with a new AI-themed twist.

This is going to sound harsh, but if your job can be automated, you have no right to expect to continue to be paid to do it.

[6741009736509Pacmikey]

.

[67410097365e3Pacmikey]

.

[67410097367bdfox]

Just a thought on automation: it hasn't taken all those jobs yet because most certainly it's still cheaper and/or more convenient to exploit "human capital" in many cases. As soon as that changes and employers can come up with the investment to setup their workplaces for it, the vast majority of them will go for it, even if they are smart enough to know that drastic societal shifts could backfire on their business model in the future. Believing in ulteriour motives of any industry would be just naive. And since nobody can realistically expect them to voluntarily share any of their profits with anybody either, governments need to prevent that development by regulating the crap out of it. And they could if they wanted to.

For AIs that could turn into a very different story very fast. Not only is it far less hindered by the need for high investments and manufacturing problems but it is also reasonable to think that there is a point of no return coming up, probably sooner than most expect. Then the cat is out of the bag indeed. These are things you can't wait for them to happen because you personally lack the foresight before you can accept the fact that the potential for catastrophic consequences is simply too high to not take it serious before shit hits fan.

And yeah, seems almost too corny to bring up because it's such a SciFi-cliche already -cough cough- but the combination of rogue AIs with a highly automated world might turn out a bit too real some hot winter day.

[6741009736891Pacmikey]

.

[6741009736cd9Nameless Voice]

Nameless VoiceJust imagine how much better our society would be if we got AI to replace all the white collar bullshit jobs. So, so many people make 100-200K basically doing nothing. A university I went to, has the "President" getting paid MORE MONEY than the PRIME MINISTER OF CANADA. And of course, the students are the ones paying that fat fuck's salary. That's not even counting the hundreds of "Administrators" of bullshit. These are the exact people I want the AI wave to [REDACTED]

That's not how it ever works though.
The people doing the actual work get automated away, a new group who can control/maintain the automation get brought in at higher pay, and the majority of the money goes to either the shareholders or the executives.

So you end up with a further divided society - more people become poor, while the rich get richer.


As always the problem isn't automation or AI, it's our society.  So long as we believe that everyone should be forced to work (and, in most cases, be heavily exploited) in order to be allowed to live, automating away jobs will always end up making things worse for many people, rather than freeing them up for higher pursuits like it's supposed to.

[6741009736dbePacmikey]

.

[6741009736ee1voodoo47]

yes yes, the world is garbage, but if I had to choose between a miserable life today, and a miserable life 100 years ago, then I would take now any day. compared to what is a very important question.

remember, the choice is not hell or utopia, but hell or a thing somewhat less of a hell.

[6741009736facPacmikey]

.