In this post, I’ll introduce a reinforcement learning (RL) algorithm based on an “alternative” paradigm: divide and conquer. Unlike traditional methods, this algorithm is not based on temporal difference (TD) learning (which has scalability challenges), and scales well to long-horizon tasks.
We ...
#484 – Dan Houser: GTA, Red Dead Redemption, Rockstar, Absurd & Future of Gaming
Dan Houser is co-founder of Rockstar Games and is a legendary creative mind behind Grand Theft Auto (GTA) and Red Dead Redemption series of video games. Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep484-sc See below for timestamps, transcript, and to give feedba...
Re:imagine 2025 Dallas - Where AI Velocity Meets Control
The Dallas stop of re:imagine 2025 opened with proof instead of prediction. A customer dialogue withAMD’s Kari Ramseth, HR Service Center Leader, illustrated what disciplined scale looks like in practice.
Post-Training Generative Recommenders with Advantage-Weighted Supervised Finetuning
Author: Keertana Chidambaram, Qiuling Xu, Ko-Jen Hsiao, Moumita Bhattacharya(*The work was done when Keertana interned at Netflix.)IntroductionThis blog focuses on post-training generative recommender systems. Generative recommenders (GRs) represent a new paradigm in the field of recommendation syst...
Protecting What Powers Progress – A Modern Look at Data Security
Modern progress runs on information. Every business, no matter the size or industry, depends on the constant movement of data to function, serve customers, and grow. The more digital the world becomes, the more vital it is to protect that information. Data security defines a company’s reputation, re...
AI Startups and the Legal Risk of Getting It Wrong
Artificial intelligence (AI) is exploding. There were 5,509 AI startups in the US between 2013 and 2023. And according to Statista, they’re receiving a massive amount of funding. “ In 2024, AI startups received more than $0.5 trillion and raised over $100 billion. “ In 2024, AI startups received mor...
#483 – Julia Shaw: Criminal Psychology of Murder, Serial Killers, Memory & Sex
Julia Shaw is a criminal psychologist and author who in her books explores human nature, including psychopathy, violent crime, the psychology of evil, police interrogation, false memory manipulation, deception detection, and human sexuality. Thank you for listening ❤ Check out our sponsors: https://...
The Hype Trap: Why Waiting for Certainty Guarantees You’ll Be Too Late
"Doubt is not a pleasant condition, but certainty is absurd." — Voltaire
Voltaire nailed something every business leader knows instinctively: if you wait for certainty, you’re waiting for an illusion. In business, certainty is just another word for “too late”—the point where opportunity has been cl...
Designing Transformative AI Experiences: Lessons from Hash Bajwa
"The most powerful technology doesn’t just solve problems. It inspires people to see what’s possible," says Hash Bajwa.
In Episode 6 of AI Activations, Bajwa reflects on a career spanning the United Nations, Apple, and the Obama Foundation. He shares how organizations can design AI experiences that...
AI News Weekly - Issue #456: How AI could radically change schools by 2050 - Sep 19th 2025
In the News
How AI could radically change schools by 2050
By 2050, education will look radically different than it does now, according to psychologist and social scientist Howard Gardner — the originator of the theory of multiple intelligences — and Anthea Roberts, a visiting...
AI News Weekly - Issue #455: Albania appoints world’s first AI-made minister - Sep 11th 2025
Powered by
constellationenergy.com
In the News
Albania appoints world’s first AI-made minister
Albania has become the first country in the world to have an AI minister — not a minister for AI, but a virtual minister made of ...
What exactly does word2vec learn, and how? Answering this question amounts to understanding representation learning in a minimal yet interesting language modeling task. Despite the fact that word2vec is a well-known precursor to modern language models, for many years, researchers lacked a quantitati...
Whole-Body Conditioned Egocentric Video Prediction
×
Predicting Ego-centric Video from human Actions (PEVA). Given past video frames and an action specifying a desired change in 3D pose, PEVA predicts the next video frame. Our results show that, given the first frame and a sequence of actions, our model can generate videos of...
"In projecting language back as the model for thought, we lose sight of the tacit embodied understanding that undergirds our intelligence." –Terry WinogradThe recent successes of generative AI models have convinced some that AGI is imminent. While these models appear to capture the essence of human...
Pope Leo XIV Declares AI a Threat to Human Dignity and Workers’ Rights
Pope Leo XIV is taking a bold stance on artificial intelligence, calling it “a challenge to human dignity, justice and labour” in his first major address since being elected leader of the Catholic Church. The new pontiff is placing AI at the center of the Church’s moral agenda, warning that we’re en...
AI May Soon Help You Understand What Your Pet Is Trying to Say
Chinese tech powerhouse Baidu has filed a patent for a system that could use AI to decode animal sounds and behaviour then translate those signals into human language. For the millions of pet owners wondering what their animals are thinking, this could be the first real step toward bridging the comm...
Murder Victim Speaks from the Grave in Courtroom Through AI
Chris Pelkey was shot and killed in a road rage incident. At his killer’s sentencing, he forgave the man via AI. In a historic first for Arizona, and possibly the U.S., artificial intelligence was used in court to let a murder victim deliver his own victim impact statement. What happened Pelkey, a 3...
Defending against Prompt Injection with Structured Queries (StruQ) and Preference Optimization (SecAlign)
Recent advances in Large Language Models (LLMs) enable exciting LLM-integrated applications. However, as LLMs have improved, so have the attacks against them. Prompt injection attack is listed as the #1 threat by OWASP to LLM-integrated applications, where an LLM input contains a trusted prompt (ins...
Repurposing Protein Folding Models for Generation with Latent Diffusion
PLAID is a multimodal generative model that simultaneously generates protein 1D sequence and 3D structure, by learning the latent space of protein folding models.
The awarding of the 2024 Nobel Prize to AlphaFold2 marks an important moment of recognition for the of AI role in biology. What comes n...
The CxD Interview GuideGet the most out of your conversation design job interviewInterviews go both ways. In these trying times of layoffs and non-existent UX job listings, it might seem a little pretentious to ask that candidates be “more picky” about their future employer, but the questions you po...
This Week in AI: Meetups, Courses, and Community NewsWelcome to this week’s edition of ChatBots Life Update, where we aim to provide you with the most crucial AI information in just 5 minutes.Here is what’s in store:💪 Upcoming Online Meetup: Join us and Say Hi🎁 Big Content Update: Courses, Deep Dive...