Elon's biggest problem @ Twitter - he's not funny at tweeting

Yeah, I don't think Twitter's data is worth 70B (Reddits eventual market cap might serve to be the best proxy/guestimate) but it's not insignificant for training AI.

A lot has already been mined before, so the next step might be what LLMs train from its own generated data. That plus more efficient and accurate video inference and training (Gemini's demo of that ideal was admittedly yugely misleading), which isn't X's forte.

 

Where does the 70B come from?

Hollow valuations come from wowzer demo's like Gemini. ChatGPT 3.0 was one of the first models to demonstrate the power of LLM's, however at the end of the day, its still just a wowzer demo. It primarily solved the tedious task of someone to mine their own data (ie: a research paper). Its still lacking because its not accurate. Its accurate enough to submit for a homework assignment but not accurate enough for someones life to depend on it.

At the end of the day, its about moving data efficiently. Native Indians used smoke signals. In the mid 1800's, we discovered the ability to send analogue electrical signals across wire (morse code). Early 1900's we discovered the ability to transmit and receive radio frequencies. Fiber has allowed us to transfer massive amounts of data near instantaneously. Now we have the problem of too much data for a human to interpret.

Then we should ask what the value is in each of the large data repositories (google, meta, apple, amazon, X, ect ...). Googles data is largely collected, stored and optimized for advertisers. Very valuable for marketers and the reward stimulus part of psychology, but not that great at understanding the underlining fundamentals of humans and the universe. Reddit data is much better at this. Meta is just dying. The next Yahoo. Who really knows whats happening at Amazon. X is in the transformation stage. 2024 is the make it/break it for X. If there is a massive faux backlash against X over the 2024 election, I think it will hit a Make It point and its data becomes extremely valuable.

 

2023 was supposed to be it, as per Musk rationale for paying such a high price for it.

as it turned out, in 2023, Elon's hubris broke X's advertising revenue. ​

 

Not looking them up, so I could be wrong here, but I think...

- China probably has the largest video surveillance dataset, and they also don't have to worry about user privacy
- The largest dataset is the web as a whole
- The largest datasets owned by private companies (ignoring privacy laws) include Google (Google Search, YouTube, ...), Meta (Facebook, Instagram, WhatsApp, ...), TikTok, and so on...

It's not just about the dataset but also about the type of data (e.g., YouTube 'video' vs. Tesla 'miles') and its quality. An LLM trained on social media data is something I wouldn't trust very much. An LLM is "biased" based on the dataset it is trained on, and social media can be quite biased depending on the platform, each with its niches and biases.

I don't think xAI has any advantage on the dataset side (in fact, it probably has a disadvantage if each company lock down their own dataset and they can somehow ignore privacy laws). I believe you are putting too much stock in that. However, there are other areas of innovation needed to make the next leap; it's not just about the data.

 

I meant the 40B purchase price Elon bought it for.

As for the rest, there's no indication that X/Elon is ahead of the curve in any LLM tech. Semi-Autonomous driving uses something else and maybe his slavebots can leverage some LLM models, but prob not next level like...hiring a dancing clown in spandex to woswza the shareholders into thinking this will be the final product

 

Last month, they claimed to be behind ChatGPT-4 but ahead of ChatGPT-3.5 and Claude-2 in some metric (actually, just a single important metric). Of course, it's not just about being technically better but also about having a customer base. Amiga was well ahead of its time but succumbed to the inferior IBM x86 architecture. OpenAI is well-positioned, especially with Microsoft's investment and partnership. xAI has a long way to go to compete.

ps. Personally, I prefer Claude over ChatGPT. The default setting for ChatGPT is too conservative (not politically but in terms of safety), but I haven't tried ChatGPT's custom instructions yet.

 

I also prefer Claude over chatgpt. I haven't done any "prompt engineering" work, but it's better for handling large files and explaining nuace in my experience. I guess both have improved over time for curtailing BS facts, but anthropic seems to be more focused on that front.

The amount of developer enthusiasm for OpenAI really sets it up to become a platform that isn't necessarily handcuffed to Microsoft's ambitions. At least that was before the failed coup, but that ecosystem sprouted within a year like those nature docs of a desert after a flash flood.

 

from post #852

since my trade, TSLA stock price was up to as hi as 252; this morning, on the heels of the news of
Tesla Recalls 2 Million Cars to Fix Autopilot Safety Flaws,
, the stock price is falling, trading ~~ 233.

Dec expiration date is Fri, Dec 15'; there is a good chance that this recent downward trend will drop the stock price to be below 230

 

premature ejockulation on SG' part on 28 Nov

it then proceeded to rise up to as hi as 252
​

this morning, 2 trading days before Dec expiration,

on the heels on the news that over 2 million Tesla will be recalled, it has dropped below 240,​

 

*gets popcorn*

230.9 with 2 DTE.

 

kari lake and musk... I am shocked. Shocked I tell you.

 

Top Free iPhone Apps

1. Temu: Shop Like a Billionaire
2. CapCut - Video Editor
3. Max: Stream HBO, TV, & Movies
4. Threads, an Instagram app
5. TikTok
6. Instagram
7. Google
8. YouTube: Watch, Listen, Stream
9. WhatsApp Messenger
10. Gmail - Email by Google

https://www.apple.com/newsroom/2023...-top-apps-and-games-of-2023-on-the-app-store/

 

Smashing. 24 hours more hours to go. Rooting for you buddy. Setting up a GFM you in case things go south

 

a rising tide lifts all boats

• looks like it'll closed above 250 by expiration.

​

will have to pay $20 to close my 2 synthetic shorts;
net loss on these 2 will be $9.50 (20 - 10.50)​

 

They met through their stylist.

 

Respect for posting a L, even if it is $20.00.

I'm only taking advice from @Sajan . He has never taken a L, especially with TSLA. Easiest stock to make money.

 

i lost $9.50; sold the shorts and collected $10.50 in advance, then paid $20 to buy back the synthetic shorts

it's a classic eg of rising tide (J Powell's re-assuring doving talk) lifting all boats.

 

1. Don't lump me with adoo. I have not posted a single trade I've made in this thread.

2. TSLA is up 25% in less than 2 months. It probably has more room to run to the mid 260s. I am not allergic to making money.

Your lack of reading comprehension implies you might be foreign or English is your second language. It's like you read the words but don't really understand what the other person is saying. I've worked with a few people like this.

 

Adoo is a good person. I dont mind being lumped in with him.

Im confused. This is the second time you have implied you have done well with TSLA and both time I have said you have done well and said I should follow your trading strategies