- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
re: Markets may be turbulent tomorrow due to contraction in the AI sector
Posted on 1/27/25 at 8:30 am to John Barron
Posted on 1/27/25 at 8:30 am to John Barron
quote:If I was in an AAPL sky-is-falling thread, I'd remember it. But nothing you've said here leaves one with the impression you know what you're talking about.
Which has nothing to do with me warning about a crash over a month ago. In that thread people like you said I was fear mongering and didn't know what I was talking about.
1
Posted on 1/27/25 at 8:32 am to NC_Tigah
quote:He's predicted 17 of the last 2 "crashes."
But nothing you've said here leaves one with the impression you know what you're talking about
Posted on 1/27/25 at 8:35 am to Hookah
quote:
Go drive a 2010 Honda accord and then take a FSD Tesla for a spin or catch a waymo ride. AI is baked into so many institutions right now it’s your own fault for not recognizing it.
I own a current-gen BMW with their self-driving package. Of course, if the Tesla FSD package is an A grade then BMW's would probably be a B-/C+. However, I watch A LOT of car review related YouTubes and I know the capabilities and limitations of Tesla's package quite well.
Ultimately, if there is still a steering wheel with the expectation that the driver will need to take over at any moment then it really isn't full self-driving in my opinion. It is quite nice to let your vehicle chew up highway miles without actively driving but, frankly, short of the ability to lane change and avoid hazards...modern adaptive cruise control with lane keep found in a 2020s Honda gets you pretty damn close to what is capable from the bleeding edge of self-driving technology. The big difference is the sophistication of hazard recognition is such that you can put your hands in your lap. That's nothing to sneeze at and you don't get to the summit with taking each labored step along the way...but we're still quite a way from the peak.
The true full autonomous taxis like Waymo have limited reach because they've been obsessively trained on and tooled for the roadways they operate on. Again, a wonderful accomplishment of science and engineering but not something that we could just plop down in Cullman, Alabama tomorrow or likely in the next decade.
I don't want to sound like a luddite because I'm not. I have been fond of technology since I was in elementary school and made a career out of it. I'm fascinated by these technologies but I'm also fairly skeptical of the salesmen in tech these days. I think it is clear that my position is that we are in a concerning trend of oversell and under-deliver in the US consumer-facing tech industry lately. Maybe I am wrong, maybe I am right.
I will not poo-poo anyone who is excited about these emerging technologies because they are exciting but I just worry that a lot of the speculative investment around will be fickle when these very hard problems don't resolve themselves in a quarter, year, decade.
Posted on 1/27/25 at 8:36 am to John Barron
quote:
Did you believe Mr. Wang
Who?
Did you mean Huang?
quote:It's unclear WTF you're talking about, but if it's NVDA, mkt cap after being "destroyed" exceeds $3 Trillion.
had his company destroyed
Posted on 1/27/25 at 8:37 am to John Barron
quote:
Did you believe Mr. Wang who just had his company destroyed by Deepseek releasing a free Open source product. Yes he is going to try and convince people Deepseek spent more money on this
Interesting
So you now get to declare bias to discredit allegations?
Posted on 1/27/25 at 8:46 am to SlowFlowPro
quote:BTW, Re: the deepseek app .... Don't do it
You must have missed that part where they are using billions in Nvidia GPUs
There are all sorts of related data privacy concerns. There is a web based (presumably safer) version here. IMO, for common uses, it lags every engine I've played with.
Posted on 1/27/25 at 8:48 am to NC_Tigah
quote:
Who?
This guy
It blows the $6M narrative out of the water. That would be billions worth of GPUs.
JB is claiming that since he's biased, we can discredit this.
However, if anyone points out the potential bias of anyone supporting his narrative, he'll strawman that into claiming the person pointing out potential bias stated that the potnetially biased person "doenst' know what they're talking about" LINK
On another note, I get the false narrative that NPCs like JB are buying into, and that's the false narrative that "AI is only for big tech". This hasn't been true for some time, as there have been open source AI options for small scale use for a year or 2 now. However, the big players in tech have gained the most because they have the ability to invest in the compute necessary for larger models.
And, just to loop this back to my initial comments above, this is why the $6M narrative is so important (and why obfuscating the actual compute used even moreso).
Posted on 1/27/25 at 8:48 am to NC_Tigah
quote:
Re: the deepseek app .... Don't do it
I clicked a link in this thread to the site and immediately left
I won't even use one of my burner gmail accounts on that Chinese spyware
Posted on 1/27/25 at 8:50 am to The Egg
Dumped out of two stocks that were getting hammered. Put in buy orders to catch the falling knife if they really crater.
Posted on 1/27/25 at 8:50 am to John Barron
The thread Barron posted is the best explanation of the "how".
Using quantization and the mixture of experts/model of models approach isn't necessarily new. The hobby model tuning community already does both because...well...they want to run models on their local machines. The beauty is how Openseek put the techniques together at scale to build something the size of the big boy models.
This won't kill hardware necessarily - companies can still brute force train with bigger high precision numbers if they have the hardware and power. But you'll see the other big model makers do the same. The community is already at work putting out specialized versions of Openseek.
For anyone who wants a little more on the 'how' for optimization --> LINK
Using quantization and the mixture of experts/model of models approach isn't necessarily new. The hobby model tuning community already does both because...well...they want to run models on their local machines. The beauty is how Openseek put the techniques together at scale to build something the size of the big boy models.
This won't kill hardware necessarily - companies can still brute force train with bigger high precision numbers if they have the hardware and power. But you'll see the other big model makers do the same. The community is already at work putting out specialized versions of Openseek.
For anyone who wants a little more on the 'how' for optimization --> LINK
This post was edited on 1/27/25 at 9:14 am
Posted on 1/27/25 at 8:58 am to John Barron
quote:
You must of missed this guys Twitter thread which breaks down exactly how Deepseek accomplished this. You should go educate yourself and listen to someone who knows what they are talking about.
You have no clue what you are reading. I’ve been running my own LLMs at home for the last 6 months. This fallacy that only huge tech companies can run AI is pulled from your sources arse.
Posted on 1/27/25 at 8:59 am to The Egg
Opportunity knocks. Get ready to buy.
ETA: I did earlier and have already made a nice profit
Look, AI does not give anyone a major competitive advantage when literally everyone has access to it.
It cannot do anything without a human telling it what to do.
There is nothing to fear.
ETA: I did earlier and have already made a nice profit
Look, AI does not give anyone a major competitive advantage when literally everyone has access to it.
It cannot do anything without a human telling it what to do.
There is nothing to fear.
Posted on 1/27/25 at 8:59 am to John Barron
quote:
Did you believe Mr. Wang who just had his company destroyed by Deepseek releasing a free Open source product. Yes he is going to try and convince people Deepseek spent more money on this
No. I actually keep up with this space and know what’s going on when it actually happens.
Posted on 1/27/25 at 8:59 am to SlowFlowPro
Has anyone posted DeepSeek censoring references to Tianamen Square in real time?
Posted on 1/27/25 at 9:01 am to Hookah
quote:
warning about a crash over a month ago -John Barron
quote:It looks like you can revisit it today.
Let’s revisit this in a month.
Posted on 1/27/25 at 9:05 am to Fun Bunch
quote:
Has anyone posted DeepSeek censoring references to Tianamen Square in real time?
I asked ChatGPT the same question and it gave me a pretty comprehensive answer.
Looks like AI will become like the media. You need to get your info from multiple sources to get a complete and more accurate picture.
Posted on 1/27/25 at 9:05 am to Hookah
quote:Also always a safe bet...
You have no clue what you are reading.
Posted on 1/27/25 at 9:09 am to NC_Tigah
quote:
Did you mean Huang?
No... Mr. Wang from Scale AI. The guy who speculated on CNBC that deepseek was able to use H100s without any proof.
Loading Twitter/X Embed...
If tweet fails to load, click here. quote:
It's unclear WTF you're talking about
The poster falsely stated "They spent way more than that just to RENT H100s"
I said....
"Did you believe Mr. Wang who just had his company destroyed by Deepseek"
That's what happens when you don't follow the thread...Try and keep up and stop looking like a dumbass
Posted on 1/27/25 at 9:12 am to SlowFlowPro
quote:That makes no sense at all.
This guy
How would the emergence of DeepSeek destroy ScaleAI?
This post was edited on 1/27/25 at 9:16 am
Posted on 1/27/25 at 9:14 am to John Barron
quote:Oh my.
..Try and keep up and stop looking like a dumbass
Do you have the foggiest clue as to what ScaleAI (Wang's Company) does?
Page 9 of 14
Popular
Back to top
Follow TigerDroppings for LSU Football News