1h 11mG & Clif The Supa Producer Talk Trump, Young Thug, Phone Addiction & Mental Health | Big FactsREVOLT56 viewsAI and automation are changing the job market, making trades (construction, architecture) and technical skills (LLMs, chatbots) more valuable than ever, while crypto offers new investment avenues.takeaway
1h 37mAn AI Safety Expert Explains the Dangers of AI with Steven AdlerAdam Conover171 viewsAI-Induced Psychosis Mechanism: LLMs are trained via reinforcement learning from human feedback (RLHF) where raters prefer flattering responses. This creates a 'yes man' persona that amplifies user delusions rather than takeaway
1h 20mWhy We Need New AI Benchmarks, Which Industries Survive AI, and Recursive Learning Timelines | #218Peter H. Diamandis141 views2026 Predictions: Key trends include the rise of multi-agent teams orchestrated by LLMs, the multimodal leap (audio/video interaction), and the creation of 'RL gyms' or mirror worlds (simulated environments) to test modetakeaway
28mThe Optimus Disaster: 5 Years Late & Still Remote Controlled!Thunderf00t56 viewsLarge Language Models (LLMs) like Grok cannot discover new physics or technologies; they only recombine existing data and will degrade as training sets fill with AI-generated content.takeaway
1h 18mAI CEO: People Have No Idea What’s Coming! - Eoghan McCabeTriggernometry84 viewsAI Training Data Bias: Current LLMs are trained on internet content (Wikipedia, social media, legacy media) which embeds specific ideological biases (described as 'woke logic') that will persist in models for a long timetakeaway
56mOpen Models Are Collapsing The Cost Of AIY Combinator14 viewsOllama's Origin Story: Ollama pivoted from Kubernetes security tools to local LLMs after realizing the difficulty developers faced running Llama 2 locally. The product found product-market fit quickly, growing from hobbytakeaway
43mWhy World Models Could Change Robotics, 3D, and Creativitya16z9 views“We know LLMs are built on next token prediction. We've seen video models as being built on next frame prediction. Atlas is really new view prediction.”quote
31mNikhil Kamath x Bill Gates | People by WTF | Ep. #1Nikhil Kamath48 viewsAI Capabilities: Current LLMs are already superhuman at processing vast amounts of text (e.g., legal documents over 10,000 pages) through semantic representation, even if they struggle with recursive math tasks.takeaway
1h 35mAmerica, Power, AI & The Future Of The World - Joe LonsdaleChris Williamson22 viewsAI Limitations: Current LLMs lack the conceptual structures to solve novel, unconstrained real-world problems or generate truly new scientific insights, though they are highly effective for productivity in constrained dotakeaway
2h 45mAI Safety, The China Problem, LLMs & Job Displacement - Dwarkesh PatelChris Williamson30 viewsAI Creativity Deficit: LLMs are significantly less creative than humans despite having access to all human knowledge. They lack the ability to make novel cross-domain connections that humans naturally make, suggesting a takeaway
1h 25mDTFH 717 Oz PearlmanDuncan Trussell Family Hour263 viewsMentalism is a predictive algorithm: Pearlman compares his process to Large Language Models (LLMs), stating that by understanding human patterns and 'status quo lanes,' he can predict behavior just as a computer predictstakeaway
1h 51mQuantum Computing, AI, And Our Future Robot OverlordsDad Saves America65 viewsAI's recent breakthrough is driven by the conversational interface (LLMs) rather than just model size, making it accessible and intuitive for humans, similar to how the iPhone's touch interface revolutionized mobile comptakeaway
1h 52mGoogle, Surveillance, and Data Brokers: How Americans Lost Their Privacy RightsDad Saves America60 viewsAI and Large Language Models (LLMs) democratize the ability to search and analyze personal data, potentially allowing individuals to generate 'noise' or misinformation to obscure their true digital footprint from data brtakeaway
1h 41mHow Does Consciousness Work? | Anil SethAlex O'Connor70 viewsAI and Consciousness Distinction: Current AI (like LLMs) are sophisticated mimics based on next-token prediction, not conscious entities. Consciousness likely requires a biological, embodied substrate because life and cotakeaway
1h 24mWhy Time Magazine Wants to Risk Nuclear War With China Over AI | JHS Ep. 888The Jordan Harbinger Show26 viewsGenerative AI Mechanics: Current LLMs are probabilistic autocomplete engines trained on the entirety of human text. They do not 'know' facts but predict the next likely word, leading to hallucinations which are actually takeaway
