Izašla je verzija 1.5 Groka (AI sistema firme Tesla Elona Maska). Veliki napredak u odnosu na prvu verziju.
Par zanimljivih komentara ispod Youtube videa.
* Hang on, wasn't Grok launched in December, 4 moths ago?
And now "it finally happened, massive update." It seems to me the acceleration is visibly becoming hyperbolic at this point.
I think AGI will come much, much faster than anyone predicts. We're not two years out anymore.
* * That s the KEY point : this news shows the pace of AI development .. probably across the whole space, not just at X/Tesla.
* * Musk thinks certainly within 5 years and most likely 2025. It seems to be the consensus view. I guess it will be unless some unexpected constraint emerges at the last moment.
* i hope grok succeed and will not go closed source
* That robot looks pissed off.
* You do realise Grok is part of the Tesla Company AI research don't you? Telsa if a half a TRILLION dollar business whereas X was valued around $50 billion.
* Dojo can train AI's on all that real world data the cars have been collecting, no one else has that much real world/human data. By comparing real world with simulated world they can make better simulated worlds. Using that for those robots and you can expect Elon to surpass almost anyone and also apply it to manufacturing - Car, Starship, Starlink, Robots, Solar Cities.........
* Meta accidentally leaked Llama, thought "Oh shit we made a mistake", then found out that it was good business for some reason I don't understand. After that they spent mind-blowing amounts of money on hardware to train Llama 3 and is going for open source AGI. Now it's a genuine possibility that X will go toe-to-toe against Meta to be the king of open source AI.
These two titans might fiercely compete for AI supremacy. I believe Meta has the compute and talent, but it will be interesting to see X at least try to beat them. I believe the best business model will be to open source models and sell cloud compute for fine-tuning and inference of these beasts. I think that is going to beat the proprietary models long term because cloud is now profitable beyond comprehension.
*
Hmm.. One should first understand that philosophically, there is no difference between a model which is actually sentient to a model which acts indistiguishibly from being sentient. Such models are philosophically sentient. If such a model is released as open source, it could mark the end.
* Well that escalated fast
