Videos

'Character is something that you don't have a valuation for in your models, and our models, if you compare them to Claude, will not have as consistent of a character. Especially for these models that are going to many users, character is important.' #shorts

AI2's open post-training recipes under the Tulu brand aim to redefine modern AI training, moving beyond simple evaluations to advanced data curation. #shorts

'In the open, we have the benefit of training on outputs from very strong models like GPT-4' - discover how open recipes are changing the AI landscape by leveraging advanced models and shortcuts. #shorts

'There are definitely things that we want humans giving the answer on, but there are a lot of mechanical tasks that we can outsource to LLMs' - exploring the balance between human and AI decision-making #shorts

In this episode of The Cognitive Revolution, we dive deep into frontier post-training techniques for large language models with Nathan Lambert from the Allen Institute for AI. Nathan discusses the groundbreaking Tulu 3 release, which matches Meta's post-training performance using the LlAMA base model. We explore supervised fine-tuning, preference-based reinforcement…

'Machines of Love and Grace is essentially portraying a world in which you have the AGI that should ASI very quickly. But that scenario is bizarre,' - illustrates the unprecedented capabilities of applying AGI in a directed, isolated way #shorts

'The marginal returns are declining,' he explains, emphasizing the need for new approaches beyond just adding more compute and data. #shorts

'Scaling the right thing matters more now than ever' - OpenAI's new era of wonder and discovery in AI #shorts

'If China wants to knock out TSMC, it can' - in this clip, experts discuss the looming threat posed by China's military ambitions towards Taiwan and its potential impact on global chip manufacturing. #shorts

'We don't see how you stop more states from adopting the worst possible kind of AI regulation unless you get to something better' - exploring the impacts of recent legislative changes on AI regulations #shorts

Unknown speaker says China won't risk global economy stability over Taiwan, but smaller strategic moves are not off the table. Understanding AI strategies is crucial for future global stability. #shorts

In this episode of The Cognitive Revolution, Nathan welcomes back Zvi Mowshowitz for an in-depth discussion on the latest developments in AI over the past six months. They explore Ilya's new superintelligence-focused startup, analyze OpenAI's O1 model, and debate the impact of Claude's computer use capabilities. The conversation covers emerging…

'AI companies are shifting to a more compartmentalized structure for security' - exploring the balance between public feedback and internal security #shorts #openai

Daniel, former OpenAI policy researcher, says 'We haven't actually solved the technical alignment problem. If our training runs succeeded, we wouldn't control it.' #shorts

Dean from High Agency discusses how top employees in companies really matter and the bets on their integrity #shorts #openai

'In 2026 or so, the internet will be balkanized into the Chinese, Russian, and various Western internets. Is this already happening?' #shorts

'In four years, you won't have access to AGI agents as they'll be controlled by a 'government, corporate, military, racing, AGI complex,' profoundly changing politics and power dynamics.' - Revolutionary discussion on the future of AI accessibility. #shorts

In this episode of The Cognitive Revolution, Nathan explores AI forecasting and AGI Lab oversight with Dean W. Ball and Daniel Kokotajlo. They discuss four proposed requirements for frontier AI developers, focusing on transparency and whistleblower protections. Daniel shares insights from his experience at OpenAI, while Dean offers his perspective…

'We don't know what weapons WWIII will be fought with, but we know we'll be back to sticks and stones' - Exploring the potential of an AI arms race #shorts

'AGI definitely exists. We are a form of AGI. I would call us a weak AGI. I don't think that is going to age very well' - expert in AI risks #shorts

In this special episode of The Cognitive Revolution, Nathan shares his thoughts on the upcoming election and its potential impact on AI development. He explores the AI-forward cases for Trump, featuring an interview with Joshua Steinman. Nathan outlines his reasons for not supporting Trump, focusing on US-China relations, leadership approach,…

'The US is uniquely ossified because our system of checks and balances makes it very hard to do more dramatic kinds of change' - Impact of political structure on innovation and competition with China #shorts

'Kamala Harris wants to cap drug prices based on international prices. This would be a huge mistake - it could delay or kill up to 170 new drugs' - the impact of drug and AI policy can't be overstated #shorts

'How do we make sure we don't have nuclear war?' EA thought leaders discuss critical tail risks and preparedness for future disasters, including AI, pandemics, and more #shorts