Well, you can always run a local model on your own hardware. And there has been progress, it isn't terrible. Personally I tried Llama with some models. But as my machine is a dev machine without much graphics card power, I only could try a highly quantized model which was working not that great. Still, I was impressed, because it was shrunk to a hundreth of the paramters, which is incredible. Would you or I even be able to form sentences if only 1% of our brain works? I doubt. Anyways, with decent hardware (and Apples stuff works well), you can run decently good local models and avoid a lot of the problems you notice.

https://www.youtube.com/watch?v=caJUD2c3QRQ

I'm not going to argue ai, but what the two of you have suggested is marksman fallacy. Also Amazon was actually profitable pretty early on in its business life it just choose to reinvest all it's earnings into the cloud and retail. As for uber they operated in a zero interest environment and thus faced lower financing costs then the current AI companies. Their not good examples either as Uber raised $20 bill and Amazon lost $5billion which is much less then most AI companies.

But I won't, because I still don't trust AI.

How much deficit do the other AI companies have?

OK, you may be right on Amazon, i just remember back then reading all the time how Amazon grew their losses year over year, and then in crash they suddenly started to become profitable. But it makes sense: they were investing heavily and therefore were in the red, but in the changed market environment of the crash they decided to become more fiscally conservative.

I had to look up marksman fallacy and was directed by search to Texas sharpshooter fallacy, which is cherry-piccking examples to fit your story. You maybe right. But man, I think that even after the hype is corrected we have *some* real use cases. As a programmer I despise vibe coding, because no one can understand and maintain that code base. But even I have to admit, that in some use cases it doesn't matter: you want a thing to work now and don't care for later. I watched a video about the math thing that was mentioned earlier, by a mathematician. I cannot judge the math, but in the middle he dropped quickly, that all the animations he used to explain the problem discussed he vibe coded. A quick program to create the animations he wanted for the video and then discard the code. No need to maintain anything. That is a useful way to go about it. I guess were is more.

Sadly I also think the governments will keep using AI for military purposes. I think that genie is out of the bottle. And from the cold war we know governments can pump out a lot of money for military uses.

So I don't know, the hype is too big for sure, these crazy investments are just that... crazy. But I think once the dust settles we have some applications and some viable business models. How big this AI business will be after correction? I really don't know. But as I assume there will be some market, I assume that some of the companies will survive. Again, probably making things cheaper and fleecce customers more.

100%.  It shocks me that people don't realize this.  Amazing bleed money for over a decade.  Apple almost went under more than once.  Nvidia was a small player for decades.  

Now all 3 are top 5 companies on the planet.

As I explained previously, this is not an apples to apples comparison; the likes of Apple, Amazon, and Nvidia always had a viable, functional business model and a clear path to profitability. By contrast, Open AI for instance doesn't even have a plan for how to become profitable: https://www.startupbell.net/post/sam-altman-told-investors-bluntly-he-had-no-plans-on-how-to-generate-revenue

According to JP Morgan, to achieve a 10% return on capital expenditure, the AI sector needs to generate about $650 Billion a year in revenue. The actual run-rate is about $25 billion. In other words, run rate is only about 3.8% what it needs to be to justify the costs.

Meanwhile, AI's contributions to the economy? "Basically zero": https://finance.yahoo.com/news/ai-contributed-basically-zero-us-181419807.html

Sure, but look man I am old-ish.  I remember when people laughed when people said everything was going to have a .com because who needed to be connected all day long?  I remember when people scoffed at the idea of mobile phones for everyone.  AI is far from the first time, in my life, where people didn't see the long-term value.  

I'm actually amazed how little Anthropic has spent on Claude compared to the other AI players, considering that Claude is considered both the best and the most versatile AI model.

Also, NVidia "AI", just some Deep learning algorithms, no real AI, unless they count the hardware sales. But then, AMD, Intel and others like Samsung anf Micron would also need to be on the list...

This. We will be giving up an invaluable technology if we act like a bunch of luddites and just ban AI altogether. I also see good points that the anti-AI people here have made. But it will be up to us to turn our electronic devices off and go touch grass in order to feel like part of the world again. We don’t have to be super connected to the rest of the world all the time. Sometimes it’s just better to feel connected to our own lives. 

also, good move on getting into quantum stocks. I’m just doing index funds, pretty boring but the expense ratio is nil and I get 10% a year still which is rare. 

People trying to stop AI or any technological advancement might as well be shaking their fist at the clouds.  It's pointless.  They aren't doing anything except excluding themselves from the process of shaping the future.  If they seriously want to improve the way the technology impacts the world then they will evaluate it impartially while acknowledging both the good and the bad.  That's how they get the respect necessary to make their opinions count.

This is why most activists fail at their supposed purposes.  They never bother to learn anything beyond surface level propaganda and people just tune them out.

I have a smart phone.  I am aware of both the positives and negatives.  I use it for work, I use it to text family and friends, but I have zero social media accounts on it.  I don't need to toss the whole thing in the trash to avoid the pitfalls.

The point of Pandora's Box is that it cannot be undone.  It's outside--now what are we going to do about it?  People who hide in the closet lose their vote.

Most of my investments are index funds but I have some speculative stuff like rare earth mining and AI energy providers (there are ETFs for those sectors, too).  These are companies that are pure revenue now but have high growth potential.  I picked up four quantum companies a few weeks ago, two of which just received money from the US government, though all of them surged.  They'll come back down again but they were investments for the future, anyway.

Which is what I think people are missing about AI.  Chatbots might not make nearly enough money to justify the expense but that isn't the point.  The point is the applications that will use AI.  It's going to be licensed out to companies who will put it in their own products then license THAT out on the enterprise level.  Will a lot of those companies fail?  Absolutely.  Will some of them be the next AWS or Salesforce?  Yeah, probably.  The financial impact of AI won't just be deep, it's also going to be ridiculously wide.