Elon Musk says xAI will retrain Grok: 'Far too much garbage
When you're Elon Musk, you don't have to rely on centuries of prevailing human understanding — you can create your own.
We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors," Musk wrote on X on Friday night.
Then, he said he would retrain Grok's latest model on that new base of knowledge to be free of proverbial waste. "Far too much garbage in any foundation model trained on uncorrected data," he added.
Musk has for years endeavored to create products, like the rebranded Twitter and Grok, that are free from what he views as harmful mainstream constraints.
https://www.msn.com/en-us/money/companies/elon-musk-says-xai-will-retrain-grok-far-too-much-garbage/ar-AA1HatEe
Overdoing the ketamine again Eloon?
= new reply since forum marked as read
And try program your "Autofill" AI to recognize what is junk and what is accurately sourced, when perhaps only 1/6 the of human knowledge is reasonably accessible on the internet.
His precious Grok will still be.trained on Auto fill slop. Just finer tuned slop, now with more harvested context and patterns to Boolean into - Is, Nots, Ands, Ors, Nors, Likes, Not Likes, and Cardinal Points.
The rest of the context of history and human knowledge still in some sort of physically written, pictorial, or aural form, or is not easily accessible in the internet if it is in a digital form, either air-gapped or firewalled, or in some arcane language or format like ADA or Fortran...
As for AI itself, as a kid from the 70's and being part of the current evolution of technology, we aren't even close to understanding how human consciousness - or even simple consciousness in any living organism evolved. For some weird reason, in the late 1990's, early 2000's, I observed that research for research sake was increasingly being dumped for research to either get products to market, or to increase "productivity". All in the name of "Biggest Bang for the Buck". A lot of promising but expensive programs that could have gotten us closer to actual AI, or understanding of how the brain functioned, was stripped and repurposed to please investors and grant administrators who wanted headlines and products to make money off of. They didn't want to pay for decades of research. They didn't want to risk failure. They wanted results.
And that's why we have datacrawling photoshoppers that are also functional spreadsheets with autocorrect/autofill like Siri, Alexa, Copilot, Gemini, ChatGPT being called "AI" instead of an actual intelligence program that can actually incorporated a form of consciousness to be able to act independently with new or unknown situations. Don't get me wrong, what's being called AI are all useful tools, true, but there's a reasoning/contextualization gap that is not yet understood. Hell, we don't even know for sure if consciousness evolved through organic chemistry or is a more mechanical quantum physics function.
We aren't there yet. And if we keep trying to tweak what we do know and have with profit based focus, we will always fall short...
