Elon Musk's xAI Unveils Grok 3: The 'Smartest AI on Earth'
Elon Musk's artificial intelligence company, xAI, has officially unveiled its latest flagship AI model, Grok 3, positioning it as the "smartest AI on Earth." This announcement marks a significant milestone in the competitive landscape of AI technology, with Grok 3 set to rival leading models such as OpenAI's GPT-4o and Google's Gemini.
Grok 3 has undergone extensive training, utilizing a massive cluster of 200,000 GPUs, which is ten times more computational power than its predecessor, Grok 2. This substantial increase in resources has resulted in a model that outperforms existing AI systems across various benchmarks.
In benchmark testing, Grok 3 has demonstrated superior performance, surpassing models like GPT-4o, Claude 3.5 Sonnet, and Gemini 2.0 Pro. Notably, Grok 3 achieved a 52% score on the AIME 2024 benchmark, 75% on GPQA Science, and 57% on LiveCodeBench. These results underscore its advanced reasoning capabilities and potential applications across diverse fields.
In addition to the model's release, xAI has introduced enhanced features within the Grok app, available on iOS and web platforms. These updates aim to provide users with a more intuitive and efficient experience, facilitating seamless interaction with the AI model.
Elon Musk's announcement of Grok 3 signifies a pivotal advancement in AI technology, reflecting xAI's commitment to pushing the boundaries of artificial intelligence and offering cutting-edge solutions to users worldwide.
Lol whoever trusts any Musk product is very naive
The tests are probably faked, or the AI is trained on test data.
Won’t touch anything coming from Musk
Good, more bandwidth for the rest of us.
Elon Musk is going to kill us all. He’s another Miles Dyson (in the movie “Terminator 2: Judgment Day”) AI is the biggest mistake that the human race has ever made. Worse by far than nuclear weapons.
Which is of course nonsense. People can only come to such absurd conclusions, if they don’t know a thing about the underlying technology and believe, that “AI” is truly “intelligent” (meaning AGI), which it is, of course, not.
AGI is in principle a real possibility, but won’t be achieved by the current approach of “just shove more data, parameters and compute into an LLM” (so multiple initially unstructured DNNs). As long as they try this, there won’t be any AGI (“real intelligence”). The only thing that is being achieved by this approach is improved 2D curve fitting, which is of course kinda nice but the results do in no way justify the means. Meaning: this is economic insanity, which is fueled by the scientific illiteracy (=stupidity) of the investors and Joe Sixpack.
I appreciate your reply, but AI as it is now is only the starting point leading in the direction of the singularity. People who can’t see that are in denial.
I have no doubt that AI will eventually be outfitted with many more capabilities, including the ability to question its own reasoning (which today’s technology only approximates). In testing ChatGTP, CoPilot, Claude, Llama3, and DeepSeek with both political, historical, and programming problems, it guesses far too much and is incorrect far too often. Sometimes it digs itself into a hole in trying to tweak an unsuccessful programming problem, missing the most obvious or simple approach. In short AI is as dumb as a rock in many ways, impressive as hell in others. While you say the “Singularity” is coming, let’s keep in mind that this notion was invented by the “Ur” tech bro of all, Ray Kurzweil. He is followed by two more generations of hucksters and salesmen of a technology that is limited by the fact that it cannot inhabit a body.