VAIBE tracks AI-focused podcasts: what changed, and the moments behind it.
← Viewing week Jun 22–28, 2026 ▾ Sep 28–Oct 4, 2026 Sep 21–27, 2026 Sep 14–20, 2026 Sep 7–13, 2026 Aug 31–Sep 6, 2026 Aug 24–30, 2026 Aug 17–23, 2026 Aug 10–16, 2026 Aug 3–9, 2026 Jul 27–Aug 2, 2026 Jul 20–26, 2026 Jul 13–19, 2026 Jul 6–12, 2026 Jun 29–Jul 5, 2026 Jun 22–28, 2026Current → Adoption Use Cases and research led the first week of tracking Across 27 episodes from 15 contributing podcasts, adoption use cases held 30.1% of the relevant conversation and research held 18.5%.
What changed The cited moments below show this week's discussion. No adjacent measured week is available for a change comparison.
Where it showed up Adoption Use Cases and Research accounted for 49% of the relevant conversation across 27 episodes.
How to read it Recurrence across podcasts does not mean agreement. The cited moments show the language behind each summary. In the discussion
Well, you know, there's a tendency to anthropomorphize it, but no, like millions of people are using Claude code because this is tangentially related to how we you know, how we come up with causal abstractions in the world. So, you know, the way we do it is we are in the world... Machine Learning Street Talk At 19:57 Hear the context This published excerpt continues in the episode. Report this quote
Overview What’s hot Keep watching
What’s hot this week Open a topic for the podcast moments behind it.
01 Adoption Use Cases Practical ways people and organizations are putting AI to work. Discussion of adoption use cases appeared across multiple podcasts for the first time. It made up 30% of the conversation across 14 podcasts this week. ML TA TC AI New this week 30% of this week · 14 podcasts → 02 Research New models, training methods, benchmarks, and discoveries from AI labs. Discussion of research appeared across multiple podcasts for the first time. It made up 19% of the conversation across 11 podcasts this week. ML TC NP AF New this week 19% of this week · 11 podcasts → 03 Safety & governance Ways to reduce harm, set limits, test behavior, and govern how AI is used. Discussion of safety & governance appeared across multiple podcasts for the first time. It made up 17% of the conversation across 13 podcasts this week. ML TC AI NP New this week 17% of this week · 13 podcasts → 04 Agents AI systems that can plan, use tools, and carry out multi-step tasks. Discussion of agents appeared across multiple podcasts for the first time. It made up 8% of the conversation across 12 podcasts this week. ML TA TC AI New this week 8% of this week · 12 podcasts → Showing 4 of 11 topics Showing all 11 topics View 7 more topics Show fewer topics 05 Economics The cost of building and running AI, and how money moves through the market. Discussion of economics appeared across multiple podcasts for the first time. It made up 7% of the conversation across 13 podcasts this week. ML TC AI NP New this week 7% of this week · 13 podcasts → 06 Regulation Laws, rules, and government decisions about how AI may be built and used. Discussion of regulation appeared across multiple podcasts for the first time. It made up 6% of the conversation across 8 podcasts this week. TA TC AF LW New this week 6% of this week · 8 podcasts → 07 Open Models AI models whose weights or code are available for others to use and adapt. Discussion of open models appeared across multiple podcasts for the first time. It made up 5% of the conversation across 7 podcasts this week. ML TA LW LS New this week 5% of this week · 7 podcasts → 08 Jobs How AI is changing work, hiring, roles, and the skills people need. Discussion of jobs appeared across multiple podcasts for the first time. It made up 4% of the conversation across 9 podcasts this week. ML TC NP TA New this week 4% of this week · 9 podcasts → 09 Hardware The chips, data centers, energy, and devices that AI depends on. Discussion of hardware appeared across multiple podcasts for the first time. It made up 2% of the conversation across 8 podcasts this week. ML TC AF TA New this week 2% of this week · 8 podcasts → 10 Infrastructure The software and systems needed to build, deploy, and monitor AI. Discussion of infrastructure appeared across multiple podcasts for the first time. It made up 2% of the conversation across 6 podcasts this week. TC AF LW LS New this week 2% of this week · 6 podcasts → 11 Other AI discussion that does not fit another recurring topic. Discussion of other appeared across multiple podcasts for the first time. It made up 1% of the conversation across 9 podcasts this week. ML TA AF LW New this week 1% of this week · 9 podcasts → 13 quoted moments 6 episodes 6 podcasts discussing this topic
What changed This was its first published appearance at 4.9% of the conversation, with evidence across 7 podcasts.
Where it showed up It appeared in episodes including “The Thermodynamic AI Computing Chip - Thomas Ahle” and “The Ad Hoc AI Licensing Regime”.
How to read it 6 cited episodes from 6 podcasts are shown below. The topic appeared across 7 podcasts in the full analysis.
Podcast moments from the week 13 quoted moments · 6 episodes · 6 podcasts 52:18 A
“it is just for a lot of things more efficient than also going through and you can get everything just the way you want it.” Open episode and seek to 52:18 Concept Open Models
Report this moment 54:21 A
“now you're really skeptical about anything because like, why would I wanna spend time reading stuff that you didn't even read yourself” Open episode and seek to 54:21 Concept Open Models
Report this moment 55:03 A
“And yeah, it's not really fun. Yeah. Have you talked with anyone about that? Like people talking about like we need like a GitHub, like a social credit system, like a karma system so people can like downrate you. And like if you don't have a high enough karma, maybe they'll ju...” Open episode and seek to 55:03 Concept Open Models
Report this moment The AI Daily Brief
The Ad Hoc AI Licensing Regime Jun 27, 2026 Open episode ↗ 05:46 A
“But interestingly, the Fable 5 background context, along with the token efficiency considerations and cost concerns that had been growing even before that, have made people a little bit less willing to just jump into the latest feature from one of the big labs. As Will Brown f...” Open episode and seek to 05:46 Concept Open Models
Report this moment Last Week in AI
#249 - Fable 5 ban, SpaceX Cursor + IPO, OSS Aplenty Jun 25, 2026 Play episode ↗ 1:06:37 A
“the cost and speed of these open source options, especially if they keep moving in that direction, will potentially start eating at the revenue of OpenAI and Anthropic” Play from 1:06:37 Concept Open Models
Report this moment 1:09:11 B
“Bottom line is, KimmyK27, Code Local. Yeah. One of the big things here is they've actually taken the, so the initial like 1 trillion parameter model and they shrunk it down to about like 325 gigs. Previously it would have been 605 gigs at full precision. So they've done this w...” Play from 1:09:11 Concept Open Models
Report this moment 1:10:28 B
“It's pretty impressive and a pretty radical quantization win here.” Play from 1:10:28 Concept Open Models
Report this moment Latent Space
Why the Frontier Ecosystem must be Open — Matei Zaharia and Reynold Xin, Databricks Jun 24, 2026 Open episode ↗ 09:12 C
“one of the reasons we open sourced it is anyone building agents, this gives them an app they can start with and customize” Open episode and seek to 09:12 Concept Open Models
Report this moment 10:05 C
“there'll be some network effect. It'll benefit from many people collaborating on it.” Open episode and seek to 10:05 Concept Open Models
Report this moment 11:42 B
“we're already seeing pull requests of all kinds of ecosystem integration” Open episode and seek to 11:42 Concept Open Models
Report this moment Training Data
Memory and Continual Learning: Engram's Dan Biderman and Jessy Lin Jun 24, 2026 Play episode ↗ 06:26 B
“Does this approach work on the frontier models or the closed frontier models?” Play from 06:26 Concept Open Models
Report this moment 06:30 C
“We need white box access to the weights, right? So we can partner with companies that have closed source weights and do this with them, but it's easiest for us to do it with open source models. But any model that's a transformer model, we can do our thing to it.” Play from 06:30 Concept Open Models
Report this moment 1:23:30 B
“I was gonna say, good luck. Let us report back with your results, please. Um, Chinese startup Z.AI released something called GLM-5.2 that's getting quite a bit of attention online. This is an open weights model under an MIT license that The Open Model Leaderboards and beats GP...” Play from 1:23:30 Concept Open Models
Report this moment
The week, in their words “ Headline evidence · Research
The analogy that I reach for here is something like a dimmer switch, where I think you can basically accommodate both the binary intuition and the sort of continuous intuition. You know, if you have a light with a dimmer switch, like, really, it is either on to some extent or ... The Cognitive Revolution ↗ 02:50 Report this quote Read 2 more voices “ Also in the conversation · Safety & governance
We do put extra effort and extra testing on things that we view as higher stakes. So one of the things that is important is we do overall testing, but, you know, we talk about your money, your life queries, right? So we put extra scrutiny on things like medical or things invol... AI Inside ↗ 49:38 Report this quote “ Also in the conversation · Agents
you have the world's most infinitely patient and knowledgeable tutor in the actual tools themselves The AI Daily Brief ↗ 11:50 Report this quote
Opening edition
This week is our starting point. Patterns need at least two weeks to compare. This is the first published week, so there is no change to show yet.
View the next edition → Get the weekly AI conversation The week’s key ideas and podcast moments, delivered to your inbox.
