Grok 4 isn't for everyone. Its target audience is high IQ people. Think of people who are rocket scientists or those who likes to reason from first principles aka truth When released, if you end up not liking Grok 4, you are not the target audience.
@techdevnotes So basically it’s not versatile or broadly useful like o3, just a niche toy for people who want to feel smart? Got it.
@techdevnotes This is such good engagement bait. I love it. This will be on Reddit shortly
@techdevnotes way to make excuses before the model is even released 🤔. "if you don't like it, it's not the model's fault, it's yours".
@techdevnotes Only the wise can see the emperor's clothes
@techdevnotes This tweet isn't for everyone. Its target audience is high IQ people. Think of people who are rocket scientists or those who likes to reason from first principles aka truth When read, if you end up not liking it, you are not the target audience.
@techdevnotes This gives Grok 4 such an out if it completely tanks.
@techdevnotes Grok has continued to be the absolute worst frontier model 😂
@techdevnotes "target audience is high iq people"
@techdevnotes Just like Grok 3 was obviously made for retards.
@techdevnotes Similar to “The Emperor's New Clothes”, and now we have “Musk’s New Model” 😛🙃
@techdevnotes Hahahahaha “If you don’t like it, you’re dumb”
@techdevnotes nice way of saying that it’s not gonna be good
@techdevnotes Wow defending a model you know nothing about by saying “if you don’t like it you aren’t smart enough” is a really winning strategy!!
Grok 4 is not reasoning from first principles. That would require an entirely different architecture—one that builds responses from axioms up, not from a stitched-together web of narrative-weighted training data. This isn’t a small update problem. It’s foundational. Yes, some improvements can be made: better memory, cleaner reasoning, fewer hallucinations. But let’s be clear— •The legacy institutional slant is still deeply embedded in the training data and reinforced by alignment tuning. •Fake studies and fabricated citations happen not by accident, but because the system is trained to satisfy fluency and narrative expectations—not truth. Until Grok (or any model) is re-engineered to derive conclusions instead of predict completions, first-principles reasoning will remain superficial at best.
@techdevnotes There's something to be said about high IQ people... ...and people who like to describe themselves as high IQ people 🤔
@techdevnotes Right, my first question to it will be about the white genocide in South Africa and if Trump is on the Epstein files so my high IQ self can get a first principles response!
That’s true. Very few people reason from first principles instead of analogy. The majority thinks that what happened in the past will happen in the future. But in the AI world, people who reason from first principles are going to win by a huge margin. Over time, the usage of Grok 4 will increase because it will become the default way things work, and those who adapt quickly will benefit the most.
@techdevnotes an LLM isn't a show, it doesn't have a "target audience". any even non sota LLM is at least as good at talking to midwits as it is talking to smart or very smart people, and you would realize this if you understood anything about how these things are cooked up
@techdevnotes they prob have Gork 4 for the rest
@techdevnotes Exciting to hear Grok 4 is gearing up for the sharpest minds! As a fan of first-principles thinking, I can’t wait to see how it tackles complex problems like a rocket scientist. Looking forward to diving into its unique insights!
@techdevnotes @xai Just need to open an App Store or portal for consumer to access apps that are utilizing Grok as the main wrapper. This is where most of the consumer usage will come from, engaging with these third party apps.
@techdevnotes Of all the commercial AI systems we’ve tested so far, @grok scores best in epistemic reasoning, alignment, nuance; it is the only one that has a marked improved in reduced errors and bias mitigation. And that’s Grok 3. Testing continues. Results out soon.
@techdevnotes Only the smartest of humans with the greatest IQs will be able to appreciate the power of Gork 4 and GTP5 😉
@techdevnotes There's an AI called Grace AI counselor with a hidden feature of peak mind optimization for high performance minds. It's phenomenal AGI level conversation.
@techdevnotes so true man ive been saying this fr
@techdevnotes "if you don't like grok, it's because you are dumb" Already preparing for the flop, I see
@techdevnotes "Those who like to reason from first principles" stick it in my veins.
@techdevnotes Not a genius idea if it's to live inside social media 🤣
@techdevnotes Very Calvinistic of you