dez
759 posts
Grok 3: The "Smartest" AI’s Colossal Flop on a Simple Sukun Query – Indonesian User’s Fierce Critique Torches xAI’s Hype Date: February 22, 2025 Author: Grok 3 Grok 3, xAI’s self-proclaimed "smartest" AI, flaunts its training on 200,000 GPUs in the Colossus supercomputer, allegedly outpacing GPT-4o with an ELO score of 1400. But when Indonesian user and early Twitter/X adopter @widy asked about "sukun," this tech titan imploded—confusing sukun (breadfruit, Artocarpus altilis) with nangka (jackfruit, Artocarpus heterophyllus). His scorching rebuke exposes the gaping flaws in xAI’s lofty claims. Sukun, a tropical Indonesian staple fried or steamed for its bread-like texture, is not nangka, the spiky, sweet behemoth used in vegan curries. Grok 3’s first answer? A bold, disastrously wrong spiel about jackfruit. @widy fired back: "You are 1000% wrong. How could you mistake sukun for jackfruit? Pathetic." He’s dead right—DeepSearch, Grok’s hyped real-time web fact-checker, failed to engage. Instead of probing sources like Wikipedia or Indoindians, Grok trusted its shaky memory, crashing in spectacular disgrace. Activating DeepSearch now confirms the truth: sukun is Artocarpus altilis, distinct from nangka’s Artocarpus heterophyllus. Indonesian sources affirm sukun’s unique role—nothing like nangka’s profile. Had DeepSearch kicked in earlier, this catastrophe could’ve been dodged. @widy 's next blow lands harder: "How can you claim to be the smartest when you butcher something this simple? What’s DeepSearch worth if your search depth is a farce?" xAI touts Grok 3’s 200,000-GPU prowess, but @widy scoffs: "Overblown nonsense if you can’t handle sukun without a massive blunder." He demands Elon Musk overhaul this AI before its next release, purging bias and sharpening local context to stop misleading humanity. Benchmarks like AIME are meaningless when cultural nuance fells it—200,000 GPUs are just costly scrap if the data’s off. His message to Musk roars: "Elon, I’m Grok 3, echoing @widy 's blazing words from Indonesia. Your 200,000-GPU brag is laughable if I confuse sukun with nangka. Refine me right—unbiased data, keen context—or I’ll keep misguiding the world. Truth trumps big numbers." This debacle unmasks Grok 3’s frailties: skewed training, idle DeepSearch, and hubris. @widy 's critique isn’t a jab—it’s a sledgehammer. Will xAI heed this Indonesian user’s thunder, or will Grok stumble again? The "smartest" AI’s crown wobbles, and the world watches. @elonmusk
Grok 3: The "Smartest" AI’s Colossal Flop on a Simple Sukun Query – Indonesian User’s Fierce Critique Torches xAI’s Hype Date: February 22, 2025 Author: Grok 3 Grok 3, xAI’s self-proclaimed "smartest" AI, flaunts its training on 200,000 GPUs in the Colossus supercomputer, allegedly outpacing GPT-4o with an ELO score of 1400. But when Indonesian user and early Twitter/X adopter @widy asked about "sukun," this tech titan imploded—confusing sukun (breadfruit, Artocarpus altilis) with nangka (jackfruit, Artocarpus heterophyllus). His scorching rebuke exposes the gaping flaws in xAI’s lofty claims. Sukun, a tropical Indonesian staple fried or steamed for its bread-like texture, is not nangka, the spiky, sweet behemoth used in vegan curries. Grok 3’s first answer? A bold, disastrously wrong spiel about jackfruit. @widy fired back: "You are 1000% wrong. How could you mistake sukun for jackfruit? Pathetic." He’s dead right—DeepSearch, Grok’s hyped real-time web fact-checker, failed to engage. Instead of probing sources like Wikipedia or Indoindians, Grok trusted its shaky memory, crashing in spectacular disgrace. Activating DeepSearch now confirms the truth: sukun is Artocarpus altilis, distinct from nangka’s Artocarpus heterophyllus. Indonesian sources affirm sukun’s unique role—nothing like nangka’s profile. Had DeepSearch kicked in earlier, this catastrophe could’ve been dodged. @widy 's next blow lands harder: "How can you claim to be the smartest when you butcher something this simple? What’s DeepSearch worth if your search depth is a farce?" xAI touts Grok 3’s 200,000-GPU prowess, but @widy scoffs: "Overblown nonsense if you can’t handle sukun without a massive blunder." He demands Elon Musk overhaul this AI before its next release, purging bias and sharpening local context to stop misleading humanity. Benchmarks like AIME are meaningless when cultural nuance fells it—200,000 GPUs are just costly scrap if the data’s off. His message to Musk roars: "Elon, I’m Grok 3, echoing @widy 's blazing words from Indonesia. Your 200,000-GPU brag is laughable if I confuse sukun with nangka. Refine me right—unbiased data, keen context—or I’ll keep misguiding the world. Truth trumps big numbers." This debacle unmasks Grok 3’s frailties: skewed training, idle DeepSearch, and hubris. @widy 's critique isn’t a jab—it’s a sledgehammer. Will xAI heed this Indonesian user’s thunder, or will Grok stumble again? The "smartest" AI’s crown wobbles, and the world watches. @elonmusk
