News Elon Musk Confirms 'TruthGPT' AI Project, Will Use Lots of GPUs

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
All this snark is amusing.
Simple truth - Elon builds the Model-T of rockets, the Falcon 9, which will fly more times in '23 than the rest of the world combined (including the U.S.), with 1st stages recovered routinely. The most reliable rocket in history, with lower prices & the most profit. And soon an 18-wheeler will replace the Model-T. He will make space available to everyone, not just astronauts.
Elon not only builds the best EV, with the best batteries, he's renovated the manufacturing of vehicles..
Elon will make a bankrupt Twitter profitable in a year, even as it gets improved. (Notice how much Twitter changed in the last 5 years? Yeah. Me neither.)
I'd be happy to receive odds on Elon's success in AI.
 
I mean, just from a business perspective: Musk himself says the company has lost 10s of billions of $ in value and is now worth half what he paid for it, in barely half a year's time. Sounds like a disaster to me.
Adds were pulled after they discovered Elon was going to allow for free speech. A lot of those companies that pulled their adds are slowly returning.
 
  • Like
Reactions: Tom Sunday
"Lots of GPU's".

So will GPU prices now go way high again as in "bitmining" times.....?

Aside from that it seems like there could be a great science fiction story here:

"Bunch of GPU based AI's asking for more GPUs to both propagate themselves and even more GPU's for bit-mining in order to fund it all."

Could be quite a number of possible story lines.

= = = =

As for good endings, science fiction or not......?

TBD.
 
His explanation of making AI safer by "trying to understand the truth of the universe" is laughably ridiculous and smacks of a snake-oil pitch.
It reminds me of the 2008 sub-prime mortgage crisis, which we learned was fueled by the "financial geniuses" making financial derivatives by bundling together lots of mortgages. They used flawed statistical models that didn't account for off-trend market conditions, when bundling together mortgages from borrowers with different credit ratings. So, when things went south, these supposedly AAA-rated securities became insolvent and many investors lost their shirts on what were pitched as rock-solid investments.

The similarity being that if you try to build a knowledge model atop a weak foundation, where facts and real truth are sparse, what you end up with is a tower of garbage. Your models, be they statistical or fancy AI models, won't save you from input that's too low-quality.

If he really cared about truth, then he should buy some newspapers and support more reporters on the ground. However, if I imagine that would play out, I'd actually rather he not.
 
Simple truth - Elon builds the Model-T of rockets, the Falcon 9, which will fly more times in '23 than the rest of the world combined (including the U.S.), with 1st stages recovered routinely. The most reliable rocket in history, with lower prices & the most profit. And soon an 18-wheeler will replace the Model-T.
Okay, so you have examples of cars and rockets where he's been successful. Maybe batteries and solar, too. So, maybe that tells us that he should stick to building physical stuff, rather than trying to play in the social networking domain.

All the time, we see examples of people attempting to transition from the music business or sports to TV and film. A lot of times it doesn't work well. The same can be true with tech.

Or, maybe his "outsider" startup mindset was just the thing needed to shake up the auto or space industries. However, social networks grew up in Silicon Valley and maybe there just aren't the same sort of opportunities for him to be a disruptor.

Elon will make a bankrupt Twitter profitable in a year,
How? He's nowhere near on track, so far. And Twitter is getting worse, not better. If a credible alternative soon emerges, it will quickly go the way of MySpace or Yahoo.