AI For Humans: Making Artificial Intelligence Fun & Practical Grok 4 is Nearing AGI But... Can Elon Get Out Of The Way?
121 snips
Jul 11, 2025 Grok 4 just passed a major AI benchmark, raising questions about its potential and Elon Musk's credibility. Meanwhile, rumors swirl around GPT-5's imminent release and Google’s DeepMind starting human trials for AI-designed drugs. The podcast dives into the emerging Comet browser as a challenge to OpenAI and discusses YouTube's crackdown on AI-generated content. With ethical AI practices in focus, they explore the fun side of human-robot interactions and even consider a future with fully AI-generated movies. Is the world ready for smarter AI?
AI Snips
Chapters
Books
Transcript
Episode notes
Grok 4 Achieves PhD-Level Knowledge
- Grok 4 achieves PhD-level academic scores, but still struggles with some simple visual reasoning tasks.
- This shows AI models can have extreme knowledge depth yet fumble easy problems humans solve effortlessly.
Grok Heavy Uses Expert Consensus
- Grok Heavy uses mixture of experts to spin up multiple reasoning models that converse and select the best answer.
- This method is costly but effectively scales AI reasoning and improves intelligence.
McKay Wrigley's Grok Heavy Demo
- McKay Wrigley paid for Grok Heavy and created impressive 3D scenes showing models forming words and emojis flawlessly.
- Other AI models fail at such dynamic code manipulation to produce human-controlled visuals.
