Kind of actually scared about the AI future...

[Gat Pelsinger]

I know, everybody has already discussed this, but nobody has really discussed a spot on and relevant answer.

 

We need no statistics that the current LLMs have already affected heavily in the job market. But the AI threat is only like 10% worse than "just helping humans". AI in the current state will not replace most of the jobs, but will help humans be more productive, which indirectly means less hiring is needed, but the current state it is not too bad.

 

But with the rate AI is progressing, it will not take long before AI gets good enough to replace us, such as with the developments of AGI (Artificial General Intelligence).

 

It is said that jobs including physical work will not be replaced by AI that immediately, which is true, but the real high paying jobs are of course programming, which is the job that will be swallowed by AI the first. Of course, human intervention will still be needed here and there, but the affect on the job market is going to be huge and this going to be like an actual crisis. Another great depression. Nobody will have any jobs.

 

And everybody is saying like the new jobs are going to emerge and all, like people who know how to use AI and all. It's all bs. AI is there so that they understand us, so there will be no reason to hire another person who controls AI. Yes, at some small points people like these will be hired, but not much. And at least I don't have any vision of what all and how many jobs because of AI will be created, and those are nothing compared to the ones that will be lost due to AI. Upon that, just how using computers have become a basic knowledge that people need to have to get hired, using AI also becomes one necessity that people need to know.

 

Of course, I am not one guy sitting here caring about the human race, and that is the reason of creating this post. What steps are being take to actually control AI development? Is the government on a close spy with AI development and making sure that nothing ground breaking gets invented and gets released to the public and destroy the world? But for the sake of science and for the other benefits though, we can't just not develop AI further, right? For that, humans need to adapt to AI, but how would that even be?

[dizmo]

People probably felt the same way about the Industrial Revolution. It's just another step of advancement. It's here, it's going to be used and experimented with, there's absolutely nothing that will change that, save a nuclear/solar event that destroys the worlds electronics. Personally I'd rather have the advancement than going backwards.

 

In the perfect world, there would be no need for currencies. Hopefully AI pushes is further towards that.

[IkeaGnome]

12 minutes ago, dizmo said:

People probably felt the same way about the Industrial Revolution. It's just another step of advancement. It's here, it's going to be used and experimented with, there's absolutely nothing that will change that, save a nuclear/solar event that destroys the worlds electronics. Personally I'd rather have the advancement than going backwards.

 

In the perfect world, there would be no need for currencies. Hopefully AI pushes is further towards that.

I'm actually very impressed so far in the Snapdragon laptops. If you're not getting one for gaming they're quite the beast. Yes, there's compatibility issues. If the stuff you use isn't on the list of "currently has issues" the seem to be great. 

If you want better compatibility OOB then the AMD AI chips seem to be a good balance. They also allow for a dedicate gpu.

 

[Gat Pelsinger]

@IkeaGnome

 

Wrong thread.

[IkeaGnome]

9 minutes ago, Gat Pelsinger said:

@IkeaGnome

 

Wrong thread.

How is it wrong thread when "AI" is getting used to close a lot of the gap?

[Gat Pelsinger]

@IkeaGnome

 

Your reply was so different and out of the topic that I thought you mistakenly replied to a wrong thread.

 

I have no idea what makes you reply about snapdragon or AMD laptops on a thread discussing about the future threats that will be caused by AI. I mean what do you mean by these things support AI? What is the connection? My brain ain't braining. Help.

 

29 minutes ago, IkeaGnome said:

How is it wrong thread when "AI" is getting used to close a lot of the gap?

What does that have to do with the post? AI helping these laptops be better is so off topic from discussing future threats of AI.

[IkeaGnome]

1 minute ago, Gat Pelsinger said:

I have no idea what makes you reply about snapdragon or AMD laptops on a thread discussing about the future threats that will be caused by AI. I mean what do you mean by these things support AI? What is the connection? My brain ain't braining. Help.

 

1 hour ago, Gat Pelsinger said:

But with the rate AI is progressing, it will not take long before AI gets good enough to replace us, such as with the developments of AGI (Artificial General Intelligence).

 

34 minutes ago, IkeaGnome said:

How is it wrong thread when "AI" is getting used to close a lot of the gap?

But you know, adding about a case where we can see AI actually getting used to close a gap that wouldn't have been able to have been closed 5 years ago isn't relevant. Has nothing to do with how fast AI is progressing. Has nothing to do with it's marketing and what it's good at. Has nothing to do 

 

1 hour ago, Gat Pelsinger said:

It is said that jobs including physical work will not be replaced by AI that immediately, which is true, but the real high paying jobs are of course programming, which is the job that will be swallowed by AI the first. Of course, human intervention will still be needed here and there, but the affect on the job market is going to be huge and this going to be like an actual crisis. Another great depression. Nobody will have any jobs.

either.

Spoiler

 

Just because it didn't revolve around doom and gloom and actually found a somewhat positive light to put on "AI" doesn't mean it isn't relevant.

[Mumintroll]

#1 would be that it could replace alot of job positions. If an AI can be as efficient as 50 humans doing the same work you can bet the companies will use it to cut costs.

 

#2 AI is a multi-purpose tool which can be used for malicious intent by criminals. For example we're already seeing fake voice records and videos created to discredit or spread propaganda.

 

#3 Large clusters of data centers draws alot of power. Those focused on AI use even more. When AI is growing more of those data centers will pop up.

 

#4 Unsupervised AI, for example self driving cars, are never foolproof. Who's to blame when such car runs over someone?

 

I do hovewer like some AI tech, like DLSS on my mediocre 4060

[emosun]

1 hour ago, Gat Pelsinger said:

humans need to adapt to AI, but how would that even be?

This message brought to you by someone typing on a flat piece of glass powered by lithium batteries with ancient Greek derived letters arranged in a seemingly random order sent via wireless transmission to a device that forwards it to a server in another Country which is then accessible via undersea cables or hundreds of space craft in orbit that beam the information anywhere on earth none of which existed over 100 years ago.

 

How will humans adapt

[Sauron]

On 8/4/2024 at 7:45 PM, Gat Pelsinger said:

But with the rate AI is progressing, it will not take long before AI gets good enough to replace us, such as with the developments of AGI (Artificial General Intelligence).

People keep repeating this but it's complete bullshit. Getting models that are orders of magnitude better than what we have would require orders of magnitude more data, and in fact there's evidence showing that the relation between additional data and additional performance is logarithmic, meaning data requirements get exponentially larger for each incremental performance improvement. Not to mention the ludicrous computing requirements and power consumption that such a model would entail. You might notice that rather than getting huge 10x improvements in chatgpt every 6 months or whatever, openai just launched other types of models (i.e. for video generation) - that's because they want to keep the hype going but they know perfectly well that they're hitting a hard ceiling. There's likely still some room for improvement with LLMs but it's not going to be this breakneck pace that they want you to expect.

 

As for AGI - we're not even close and furthermore it's absolutely impossible for an LLM to evolve into one. It just doesn't generalize like it would need to to become an AGI, no matter how big and fast you make it. It's the wrong technology for that.

On 8/4/2024 at 7:45 PM, Gat Pelsinger said:

but the real high paying jobs are of course programming, which is the job that will be swallowed by AI the first.

Lol, no it won't. Models that can produce a few hundred lines at most of code that is very unlikely to work out of the box are not going to replace programmers who write thousands of lines per day in projects with dozens of dependencies and preexisting design choices.

 

You know which jobs are currently being replaced with AI? Those of freelance bloggers who would previously be paid sub minimum wage to write listicles and clickbait trash on bottom of the barrel blogs. I don't especially mind the loss (those people are mostly students who can likely find another part time job) but I do mind the destruction of the internet through AI SEO vomit.

On 8/4/2024 at 7:45 PM, Gat Pelsinger said:

AI is there so that they understand us

...no it's not. LLMs are just stochastic parrots. They don't understand a word they write. They literally just work out what the 10 most likely next words are in a given sentence and pick one at random to give you the impression there's thought behind the answer.

[wasab]

True technological revolution will come when we reached sapient generalist ai. Now, it is overpromised and underdelivered and can at best replace jobs that can be done by a Monkey which is to say, none.