Rephonic
Artwork for The Information Bottleneck

The Information Bottleneck

Ravid Shwartz-Ziv & Allen Roush
Artificial Intelligence
Machine Learning
Openai
Large Language Models
AI Research
Generative AI
Language Models
Data Curation
World Models
Reinforcement Learning
Textai
Representation Learning
NVIDIA
Gpt-5
AI Coding
Joint Embedding Predictive Architectures
Natural Language Processing
Diffusion Models
Reasoning Models
Artificial General Intelligence

Two AI Researchers - Ravid Shwartz Ziv, and Allen Roush, discuss the latest trends, news, and research within Generative AI, LLMs, GPUs, and Cloud Systems.

PublishesTwice weeklyEpisodes59Foundeda year ago
Number of ListenersCategories
TechnologyScience

Listen to this Podcast

Artwork for The Information Bottleneck

Latest Episodes

Nathan Lambert spent three years as post-training lead at Ai2, where he built the OLMo models, and he writes Interconnects, one of the most-read technical newsletters in AI. He left Ai2 in June and is now working on a new project. He's also the autho... more

Daphne Koller wrote the book that many of us learned probabilistic graphical models from, founded Coursera, and now runs insitro, which is trying to make drug discovery a machine-learning problem.

We start with the bitter lesson. She agrees with mos... more

YouTube

In this episode, Joseph Suarez from PufferAI explains why he thinks RL never had an algorithm problem, but it had a code problem. Every part of the standard RL stack was running about a thousand times slower than it should have been, and once that go... more

YouTube

Florian Brand builds evals at Prime Intellect. The premise of the conversation is that writing a benchmark is the easy part now. Keeping the model from cheating it is the job, and it takes longer than the benchmark itself.

We get into why he thinks ... more

YouTube

Key Facts

Accepts Guests
Contact Information
Podcast Host
Number of Listeners
Find out how many people listen to this podcast per episode and each month.

Similar Podcasts

People also subscribe to these shows.

Recent Guests

Pierre-Carl Langlais
Co-founder and CEO of Pleias
Pleias
Episode: Pierre-Carl Langlais on Building Models from Data You Can Account For
Ion Stoica
Berkeley professor, co-founder of multiple startups
University of California, Berkeley
Episode: How to Turn Research Into Billion-Dollar Companies, with Ion Stoica
Jean-Francois Puget
Director and Distinguished Engineer at Nvidia
Nvidia
Episode: Kaggle Grandmasters, Agent Skills, and Why Everyone Is Overfitting with Jean-Francois Puget (NVIDIA)
Dimitris Papailiopoulos
Researcher at Microsoft Research in AI Frontiers Lab; Professor at University of Wisconsin–Madison
Microsoft Research; University of Wisconsin–Madison
Episode: AI Agents and The Golden Age of Asking Questions with Dimitris Papailiopoulos (MSR/UW-Madison)
Phillip Isola
Professor at MIT, expert in representation learning
Massachusetts Institute of Technology
Episode: Why All Models Learn the Same Thing with Phillip Isola (MIT)
Qichao Hu
Founder and CEO of Molecular Universe, part of Ses Ai
Molecular Universe / Ses Ai
Episode: AI for Science with Qichao Hu (Molecular Universe / SES AI)
Benny Chen
Co-founder of Fireworks AI
Fireworks AI
Episode: Infrastructure for AI at Scale - With Benny Chen (Fireworks AI)
Oded Rechavi
Professor at Tel Aviv University, biologist, and co-founder of QED Science
Tel Aviv University / QED Science
Episode: Broken Peer Review, AI, and Worms — with Oded Rechavi
Alex Imas
Professor at University of Chicago, expert on AI and labor markets
University of Chicago
Episode: Will AI Take Our Jobs? With Alex Imas (Google/University of Chicago)

Hosts

Ravid Shwartz Ziv
Host of The Information Bottleneck
Allen Roush
Co-host of the show

Chart Rankings

How this podcast ranks in the Apple Podcasts, Spotify and YouTube charts.

Apple Podcasts
#57
Sweden/Technology
Apple Podcasts
#93
Israel/Technology
Apple Podcasts
#162
Colombia/Technology

Talking Points

Recent interactions between the hosts and their guests.

Intelligence in an Open World - with Mengye Ren (NYU)
Q: What is the value of continual learning versus offline training, and can models learn in a truly embodied way in real-world environments?
He argues continual learning is essential for embodied agents operating in the real world due to data scarcity, non-stationarity, and the need for continual adaptation, though offline pre-training will likely remain part of the pipeline.
Intelligence in an Open World - with Mengye Ren (NYU)
Q: Do you think we will see self-improvement in the near future in AI, where models improve themselves with minimal human intervention?
Mengye explains there is no consensus yet and discusses potential fundamental limits, but notes ongoing exploration of architectures and learning strategies that could enable continual improvement without relying solely on traditional backpropagation.
What Actually Matters in AI? - with Zhuang Liu (Princeton)
Q: Do you think architecture is the primary driver of AI progress, or do data and training regimes matter more?
The guest argues that while architecture matters, the data, scale, and how we train (the interface and objectives) largely determine performance; you can achieve frontier-level results with different architectures if you get the training data and process right, and you often see diminishing returns chasing purely architectural changes without broader data and training considerations.
Reinventing AI From Scratch with Yaroslav Bulatov
Q: What is your approach to building better models while addressing energy efficiency and memory constraints?
Start from energy-aware toy tasks, design with a clear energy budget in mind, and progressively scale while measuring energy per unit of useful capability. Consider architecture shifts (like local hierarchical message passing) and embrace open, iterative experimentation rather than chasing a single benchmark.
Reinventing AI From Scratch with Yaroslav Bulatov
Q: Do you think AI coding agents will replace software engineers or researchers, or will they augment human labor?
They will augment, increasing what engineers and researchers can accomplish, but not replace them entirely. The trend is toward integrating agents into workflows to amplify productivity, with humans providing direction, evaluation, and creative problem-solving.

Audience Metrics

Listeners, social reach, demographics and more for this podcast.

Listeners per Episode
Gender Skew
Location
Interests
Professions
Age Range
Household Income
Social Media Reach

Frequently Asked Questions About The Information Bottleneck

What is The Information Bottleneck about and what kind of topics does it cover?

A tech-forward discussion series that centers on the latest research trends, practical engineering challenges, and strategic implications of generative AI, LLMs, and related infrastructures. Episodes frequently explore topics like diffusion models, world models, reinforcement of safety and alignment, and the evolving roles of AI in mathematics, healthcare, coding, and education. Guests tend to be leading researchers or industry practitioners from universities and big tech who unpack complex concepts (from formalization and verification to agentic AI and open-world learning) with hands-on examples, debates about data quality versus architecture, and considerations for production deployment. Expect rigorous technical talk blended with pragmat... more

Where can I find podcast stats for The Information Bottleneck?

Rephonic provides a wide range of podcast stats for The Information Bottleneck. We scanned the web and collated all of the information that we could find in our comprehensive podcast database. See how many people listen to The Information Bottleneck and access YouTube viewership numbers, download stats, audience demographics, chart rankings, ratings, reviews and more.

How many listeners does The Information Bottleneck get?

Rephonic provides a full set of podcast information for three million podcasts, including the number of listeners. View further listenership figures for The Information Bottleneck, including podcast download numbers and subscriber numbers, so you can make better decisions about which podcasts to sponsor or be a guest on. You will need to upgrade your account to access this premium data.

What are the audience demographics for The Information Bottleneck?

Rephonic provides comprehensive predictive audience data for The Information Bottleneck, including gender skew, age, country, political leaning, income, professions, education level, and interests. You can access these listener demographics by upgrading your account.

How many subscribers and views does The Information Bottleneck have?

To see how many followers or subscribers The Information Bottleneck has on Spotify and other platforms such as Castbox and Podcast Addict, simply upgrade your account. You'll also find viewership figures for their YouTube channel if they have one.

Which podcasts are similar to The Information Bottleneck?

These podcasts share a similar audience with The Information Bottleneck:

1. Unsupervised Learning with Jacob Effron
2. No Priors: Artificial Intelligence | Technology | Startups
3. Machine Learning Street Talk (MLST)
4. Training Data
5. Google DeepMind: The Podcast

How many episodes of The Information Bottleneck are there?

The Information Bottleneck launched a year ago and published 59 episodes to date. You can find more information about this podcast including rankings, audience demographics and engagement in our podcast database.

How do I contact The Information Bottleneck?

Our systems regularly scour the web to find email addresses and social media links for this podcast. We scanned the web and collated all of the contact information that we could find in our podcast database. But in the unlikely event that you can't find what you're looking for, our concierge service lets you request our research team to source better contacts for you.

Where can I see ratings and reviews for The Information Bottleneck?

Rephonic pulls ratings and reviews for The Information Bottleneck from multiple sources, including Spotify, Apple Podcasts, Castbox, and Podcast Addict.

View all the reviews in one place instead of visiting each platform individually and use this information to decide if a show is worth pitching or not.

How do I access podcast episode transcripts for The Information Bottleneck?

Rephonic provides full transcripts for episodes of The Information Bottleneck. Search within each transcript for your keywords, whether they be topics, brands or people, and figure out if it's worth pitching as a guest or sponsor. You can even set-up alerts to get notified when your keywords are mentioned.

What guests have appeared on The Information Bottleneck?

Recent guests on The Information Bottleneck include:

1. Pierre-Carl Langlais
2. Ion Stoica
3. Jean-Francois Puget
4. Dimitris Papailiopoulos
5. Phillip Isola
6. Qichao Hu
7. Benny Chen
8. Oded Rechavi

To view more recent guests and their details, simply upgrade your Rephonic account. You'll also get access to a typical guest profile to help you decide if the show is worth pitching.

Find and pitch the right podcasts

We help savvy brands, marketers and PR professionals to find the right podcasts for any topic or niche. Get the data and contacts you need to pitch podcasts at scale and turn listeners into customers.
Try it free for 7 days