History tells me ... eh, no comment.

Just curious about your persoective. Do you think the rate of AI advancement is accelerating in the positive direction? Do you think it has currently peaked with respect to capabilities, applications, and performance? How much do you expect these metrics will have advanced in say 5, 20, or 50 years? Trying to understand what you are saying.

 

I think it's been moving at a good pace for many years as far as the design and software goes, and now hardware is catching up making it more feasible to integrate into various products. I think a lot of it has been there before the latest hype, it's just the cool thing now to say something is powered by AI. There are many types of AI that have been in use for quite some time. Some of the examples that have been thrown out so far are narrow AI, like computer vision and natural language processing. We've been doing these things since the '90's, and advances in hardware capability over time has helped it get much better. I believe these types of use cases will still be the majority. There will be some "agentic AI" agents that tie things together in a workflow, but again, we've been doing things like that under different names (Robotic Process Automation, anyone?) for quite some time. It's cool to slap the AI badge on it now.

What's driving the latest hype is Generative AI like ChatGPT, Copilot, etc., using LLM's. It generates output based on the set of knowledge that it is trained with. The larger the model and the more knowledge it is trained with, the (allegedly) better the output will be. The larger the model, the larger the cluster needs to be (and the longer you may need to wait for an output), hence the giant, interconnected AI data centers. This is a cool toy to play with, and I'm sure there's a lot to be learned that will trickle down into improving the ability of the more narrow use cases that I believe will be much more prevalent, more centered around inference engines for specific use cases.

Consider some of the examples thrown out so far. Analyzing video and images to look for weapons or something else specific doesn't require the AI to be able to tell the difference between a giraffe and a bird and an octopus. Having that ability would be a complete waste of data points. Speech to text? How many languages? There would likely need to be separate models trained for each language, or at least limited to a couple in a single model. This is not a new capability, it's just better now due to technological advances. Is it good enough now? Probably close, and the focus will shift to reducing the cost.

There are a lot of cool use cases for large inference engines. Interconnected self-driving cars, air traffic control/monitoring, weather prediction, stock trading, etc. Anything that needs to take in a lot of data points and produce an output or make a decision quickly is a great use case for AI. On the generative side, the coding example is a good one. Click bait videos and news stories are the devil. But I quite like the squatch videos.

I surely don't think we've peaked. With all this investment in these giant AI data centers? We're in a period of exploration, with everyone looking for that new killer app. Things like ChatGPT just showcase its power, but I don't believe it's a killer app. Which workers will AI replace? We're all trying to figure that out. This exploration and experimentation gets expensive, and it needs to show some payoff. I think growth will continue for several years, but it will plateau at some point when we start narrowing down the better use cases and CEO's get tired of funding science projects that don't pay off. If I was a betting man, I'd say the peak will hit on a 5 to 10 year horizon, and we'll start seeing a pullback, with more targeted investment in areas found to have a payback. I honestly don't think it will be a bubble that pops suddenly, more a flattening of the growth curve. Beyond that, who knows?

 

My short take. I've worked as a data center engineer and project manager for about 15 years now. Ai is only just getting started and being integrated into hardware, like mentioned above. We are definitely like 5-10 years before any Ai bubble.

And then quantum computers/servers are only just getting started. My company has one. Cooling is the biggest barrier for those atm but that's also much more advanced than it was 5-10 years prior.

 

I talked with an engineer who’s worked for intel since the early 2000’s and he agreed with what’s mentioned in the previous post. AI will be part of a machine that works on a closed network to keep your private information and proprietary information secret for those who choose to keep it that way. That’s the new tech he / they have been working on.

 

I just wanna retire so I can throw my cellphone(s) in the ocean

 

I retired early about 10 years ago. The cellphone thing is real. Mine goes away after dinner. And I don't "upgrade" until the phone won't work any longer. My Galaxy 9 lasted almost six years.

It's nice to have a supercomputer in your hand, but it's even better to not need that damn thing.

Went off to college with a typewriter, graduated with a laptop. That was my technology revolution. Y'all can have the AI crap that's coming.

 

My upgrade was because the apps are no longer secure...financial apps stop working.

 

I’m gen Z so I wish my phone could’ve taken this picture but I always get the cheapest one when I need one for software updates. Mind boggling this was taken on an iPhone though from about 15 feet away


I wonder if the .mil will ever use AI for operational systems or more so for predictive models etc

 

You realize consumer tech lags military tech by about 20 years right? lol

 

Palintar

 

I think the cake is a lie. The companies pushing Ai don't even know what multi trillion dollar problem they're gonna solve for all this to be profitable. Well, neither do i, and i don't think there ever will be. But then again the finance industry and .gov are incredible at keeping a house of cards up, so i also don't think the bubble will pop near as soon as anyone thinks it will

 

I get that it may be wildly different for some of yall in IT n that jazz, but for me Ai solves literally none of my problems. Like doesn't even come close to offering me anything that i can't get just as easy in an analog way

 

Garbage stock. Lost almost $300 on it on Friday

 

i'll take that trash off your hands

 

OCTO go boom. lol

 

So Cramer said CRWV was gonna continue further down last week. I loaded in the low $90's. Planned on just selling covered calls but decided to first see how much money NVDA has backing CRWV. Turns out it's a lot. Like CRWV makes up 91% of NVDA's $4B portfolio. I think I'm gonna keep it. lol

 

Where do you stand with the R/S share count?
Today was $0.20 to $9 on the R/S 1:5 shares...

 

They did so many fucking reverse splits I ended up with 50 shares in my IRA and 20 shares in my play account. Sold all today at $27 just because I was sick of looking at it. lol

 

My current favorite garbage to dump.... Is $311 a share unreasonable?


WeLl bUy tHe dIp!

I can't dump it for $0.40 (it's most recent unusual blip high price)


I went back to see WTF I was thinking throwing that money at crap...I was flush with cash for a bunch of jackpots...including turning 2000 shares GENIUS GROUP LTD at $0.54 ea exiting 2 days later at $3.50+ and then rolled that into $UNH calls that I apparently also nailed..



HEY a55H0L3 just BUY $NVDA
it would only be a 12x ROI

 

$nbis popping +45% after hours.

News of a business deal w $msft worth up to $20B.