今日概览
4
节目总数
4
订阅源
4h50m
总时长
Top Story
【All for Today】🎙️ 每日播客精要全景图
首条总览
Y Combinator Startup Podcast
(1)
Robot-Use Agents: Why General-Purpose Models May Win in Robotics
29m
2026/09/26
📝 AI
总结
🎼One of the big surprises the last few years has been the generalizability of coding agents across different domains. And now, frontronier researchers are showing that this includes controlling robots. This LED MIt professorfess Philip Ioola to suggest in a recent viral essay that we may be entering the era of robot use agents where general purpose models could make different robots more capable. So today, Francois and I invited the founders of Wa labs and robotbo curveve. Two groups of startups that are working at the very frontier of making robots more capable with Lms..😊 Maybe you guys want to briefly introduce yourselves and just say a little bit about what each of your companies focused on. I'm Hani. I'm from model labs together with Vincent. We work on building LLM tech control robots, and we do this by doing two things. building a harness that allows LM to do this very effectively. Selecting data. and Im using that data to train better L Ms. Hi, I'm Jay, a cofounder of robot curve. We are an Eval company for physical AI.😊We measure everything, any robot, any model.Including L Ms and also classical approaches, such as a.Bs, vision language action models and word action models.And we in.Evaluate all kinds of ems like hands.嗯。Rerepars.Us.humanmanize.Cter beds, all kinds of stuff.So in the last few weeks, videos from both of you guys went quite viral on Twitter. And I think inspired, I think both of them are actually quoted in that Philip I soola essay talking about robot use agents.The videos.Showed things like LOms being able to unscrew caps, being able to communicate between multiple robots.Being able to do tasks like uncing a pen, for example.And so I thought maybe what would be fun is for Francois and I to dig in with you guys on some of the research that LED to this moment to begin. And then maybe we can also show some demonstrations of what this can actually do.😊To start, why don't we talk about some of the early research around transfer between.Language models, and.Robot control. So I want to talk about.Some of the early work on pre training language models for decision making and then also the RT2 paper, which established the VLA.So maybe Jay, do you want to tell us a little bit about some of this early work involving?Being able to use language models to do any kind of robot control, one of the.Earlier successful approaches of using AI on robots is the R T2.Paper.Where they use a pretrain language model on web.text and images.And use that to control robots.And the interesting thing about that is that it's a fine tuned version of a language model.And instead of outputting.English, for example, they just output.嗯。😊What we call an an effective pose.Which is deco coordinate that you can then translate into.Join comments.That can control the robots.In some cases, it's in some sense, it's very similar to what we see now.With flat language models.Just that instead of fine tuning.The models are good enough to just do it off the box. And this is actually quite similar to like the mapping that I have in my head.Is kind of like the COT moment.And so, like.😊When we did GSM A K and you said Suie had $8. She spent five. How much does she have now.It had to output.😊Like four hash marks and then the answer.And an E.There was no ability for it this chain of thought.And so then we allowed it to actually have like, okay, let me.Yeah。8-5 is 3.. And so like,3.As he allowed it to do this kind of thinking before it actually.Gave an action or an answer.And similarly, now, like VLlas were basically like an art.RRT2 is basically like, it has to output in action.There's no like, I can't allocate more compute for a more complex task.And then now I have this code.Chain of thought thing that I can do.And I can say, okay, even even if you were using the example where you're not actually.Output in the code you're just giving it to Astra and let it think, think, think, think, think.And an output in action.Similarly, now, like.And then with code.It's a.More homomograph complexity or lower homomograph complexity.Co code length.To just use code and really compact code to just like, here's code.Yeah, I so do you agree with you on that. I feel like.At the end of the day, it's a lot about the bitter less, right, If you give the.Agents or you give the AI model more autonomy. and you, if you unshackle it.A bit more. and give it more resources. It can actually do a lot of the things.That we find tune to do.I唔.Just gonna hop in here as well. I think the, the really interesting thing about the bit lesson here is that.😊あ。Like Vs by the sign, architecturally.They're built on top of.Language models as well, right, They, they can potentially reason They can write code.So, perhaps the.Bitol lesson here isn't necessarily what.A is actually bill necessarily, but it's what.Data is most useful.Right, like people have been.Sort of nagging at this.VL a data bottlene for years now. And we're seeing.Very, very slow progress.And so the lessona here is...(已截断)
Decoder with Nilay Patel
(1)
Can Cloudflare save the web from AI?
1h 18m
2026/09/26
📝 AI
总结
🎼Support for the show comes from E Y. AI is reshaping how enterprises operate, compete and create value, but organizations are struggling to move beyond fragmented tools in siloed AI investments to create real impact. E Y dot AI, the reimagination engine is an open, dynamic AI LED technology system at the heart of E Y by combining AI, technology and people with trusted experience. E Y.😊Organizations turn AI ambition into enterprise value because AI is only as valuable as the handset shape. Go to E Y dot AI to explore more.🎼Hello and welcome to decocoder I'm Mil La Patel, editor and chief of the Verge. And decoder is my show about big ideas and other problems. This episode is part of a twopart series on the future of business. Today, I'm talking with Matthew Prince, the CEO of Cloudfl. Cloudflare is one of the most important infrastructure companies in the world.😊It protects all kinds of apps and services from bad actors on the Internet.🎼Basically, making it possible to operate a business online. Matthew West joined us on the show about two and a half years ago. what we thought then was a wild pivot point for the Internet.But now, because of AI, it turns out things are even wilder.🎼And Matthew and Clouddflare are right at the center of.Cloud Fire found in June that bots now make up more than half of Internet traffic.A number that just keeps going up.Is more and more AI companies straight more and more of the web.🎼And now is more and more people send AI agents out to.Two things for the.どこ?CloudFare sits between websites and all those AI tools.And allows website owners some level of control.🎼They can block all those tools, allow them.Or as you'll hear Matthew describe, potentially only allow those that pay money for access. So Matthew and I talked about how to control all those spots. What kind of mess they're making of the Web.🎼And what kinds of information might be more valuable in the future as some of these payment schemes come into focus. You'll hear me ask pretty directly if some of the outcomes he's describing are actually.Good?Matthew is a thoughtful guy and his answer is something I'm still thinking about well after we had this conversation.Matthew is also at the center of another important AI debate.🎼Earlier this year, Cloudflare laid off more than 1000 people,20% of the company.And Matthew wrote an oped about that decision.Which ran in the Wall Street Journal, under the headline.🎼How I choose which Cloudflare employees to replace with AI.That is pure decoder Bay.🎼AI, big controversial decisions.And work charts all in one. So obviously, we talked about that decision in detail and what it might mean for how companies are structured in the future. One last thing before we get started, you can subscribe to decoder on YouTube where we put out new episodes every Monday and Thursday. Okayk, Matthew Prince, CEO of Cloud Flare.やか。🎼 う。うん。うん。😊うん。🎼.い。🎼う?Matt Prince, you're the cofounder and CEO of Cloudflare. Welcome back to decoder.😊🎼Thanks for having me.All right, I'm really excited to talk to you again. It's been about two years since you're on the show. I just looking back with that episode.😊And I said something you like.It's a momentous time for the Internet. And I was not even close. It's now even more momentous.Time for the internet. You are at the forefront of.Rebooting how companies work with AI that is the most decoder bait of all time.I want to talk about bots and the internet.Publishers and Google and all the things that you were in the middle of, but let's start at the very beginning, Cloudflare is it.Complicated, important company.The last time I in the show, you summed it up, you said Cloudflare is a service that makes the internet faster and protects it from bad guys.Is that still how you would describe what Cloudford does. That's it's usually when I, when I'm, and I'm trying to answer a question at a.Ttail party, and I don't want to talk to a person anymore. I say that what I'd say if I actually sort of interested in talking to the person more is.But that Kevler is trying to rebuild the Internet.The way it should have been built from the beginning, if we knew how important it was going to be.And so we run a giant network in over 350 cities worldwide, literally thousands of data centers.In all those places we have equipment running.And then we do a handful things we stop if you're trying to put an application or content online, and we make sure that it's safe.If you're a person who is an employee who's online, we make sure that.Wherever you go online and safe.We help give people.The ability to write applications is can scale to an entire Internet audience and run on the infrastructure.That we have that's the fastest growing part of our business.I think the thing that we've recently started to think about is.The business model of the Internet is changing dramatically.And, and we're sort of.Trying to help work on shaping what that future business model of the internet looks like and how can we mak...(已截断)
Modern Wisdom
(1)
How To Make Your Creativity Beat The Algorithm (Live Performance) - Dylan Gossett - #1155
1h 29m
2026/09/26
📝 AI
总结
🎼People of the UK and Ireland, I'm coming to you live. Imagine that me on stage in your city. Dublin is completely sold out, but everywhere else has got limited tickets left, and you can get yours now at Chris Williamson dot live. This is a custom built live show. I absolutely adore it. It spent over a year working on it. So if you're a fan of everything modern wisdom, Come and see me on stage this October around the UK and Ireland, Chris Williamson dot live..😊Me busy keep these hands t H something like the weather to try to unlock my mind and try to think of something clever, keep me satisfied.🎼Well take burn on the weekend, straight band 29 past the ceds and believers on rain and when to get out.がしり。🎼Compass the smell of the ball washed the.Is go on, baby.🎼Itか。Le skull.🎼我的。いえ?すまい?🎼 いべ?🎼Because I' been having trouble sleeping.Only think of me.え?🎼一生为。早あ?🎼へ?飞.🎼是我。Im headed back to Austin.🎼When I close my.我们?🎼夢?やべ?🎼えな。And they I do trying to focus towards it.あ。🎼な。When.Take simple kind of moment to put my mind at ease, it takes a early session love me to make me awake in the rain.🎼What are you believe.Im having trouble sleeping.🎼The only thing Im meeting somewhere I came.I it back to your.🎼When I could down.あ。🎼 What are you believing.🎼Ca I' been having trouble sleeping.The only thing I'm needing somewhere I came from.🎼Im headed back to Austin.When I close man.🎼That's Thank you. unbelievable. so good. Thank you very much. Great to see you.😊That one's about.Austin, Texas, obviously.Very timely.Yeah.😊Did you write most of coal in 25 minutes.Yeah, yeah. Co Co was written.OneOne of the fastest ones I've probably ever written.亮。It it started. I can kind of show you. So, it started just with.🎼呃。Maybe not this exact chord progression is a little bit different. I might have an old video of it I could show you, but.I think, at almost.I think it has started with kind of that hook the.🎼I say pressure makes diamonds head hit. You know, I was like, that's, that's great. That's a song. We got to build around it and.😊诶。No, yeah, verse by verse. And it was just.I remember I was like, had my phone type in.On top of our dog kennel in the apartment at the time and.It was like right before we went out with like some work friends.None of my friends knew I son. like at all, I didn't.I don't know, I just don't go around telling people.And Ro callle went out. and Julia, my wife was just stoked on. She was like.This is so sick.And she was just showing people like Snapchat videos of the song. She was like, you know, just.What do yall think about this? And then're like, Dlan, you sing and I was like.Cats out of the bag, I guess, and then didn't do anything with it for like.A year or two.No way. until you SAT on that. You SAT on a song that did like half a billion streams. Yeah, for your. I I wasn't it. I wasn't.I wasn't on socials like, I wasn't put anything out. I didn't do anything until we were on a camping trip.With my buddy, Caden, and his.Now, wife's M K and my wife and I Julia.And I brought my guitar and was just playing some songs. And Caden has like the best beer.If anyone I know, he's like an ANR that's not doesn't work in ANR.And he heard calling and he's like, pleaseing it again, he' likevoicing them about it.And he's like, you got to do something. You have to do something. And my wife was like, don't you should just.Post on Tiktk.And I was like, well, I'm not, you know, I'm not a Tiktoker. I was like, I almost can't stand it.Like, I don't,'s not my, it's not my thing.嗯。😊But I started doing it.And then.Coal became this.定。Whi just insane. one of my buddies I was.Best man at his wedding. Godfathered his kid, and he was playing.We ready.Friends house.Who has a karaoke machine.And I don't even know why there was a guitar there.And he'd been doing exactly the same thing. I bullied him, bullied him. I was like, no, come on, come on.Because the choices between.Him playing acoustic guitar with beautiful voice.And one of our friends sad Eo wrapping. And I was like, hey, look, the decision is really easy for me to make in terms of a pressure, The fucking floor.Dude is so low for you to get over here. Yeah, so I'm interested.What were you in life when you were doing those lyrics? I what, what is that song So I was living in Austin. I worked in Are you talking about Co, you talking Yeah, Co, I was living in Austin., I was working at the racetrack, the circuit of the Americas.Down here. So I was a.At the time, I was.And event operations, I think.I was.I was I started as an intern, I used to live there and like to.A trailer in my parents RV for months.And lived at.The job site.m,But I, uh, yeah, I was working there and.It might have just kind of been through like a tough season. Like F1 season is just kind of brutal, especially for like.The outside guys, you're just, I mean, it's just like 20 hour days that you're living out there when it's F1 time and.I don't know. It's like that. It might have been some, you know, inner interpersonal fam...(已截断)
All-In with Chamath, Jason, Sacks & Friedberg
(1)
Anthropic IPO at Risk, Meta's Muse Pop, Token Prices Fall, Open Source Gains Share, Alignment Fails
1h 34m
2026/09/26
📝 AI
总结
Al right, everybody, welcome back to episode 290 of the number one podcast in the world. It's the all in podcast with me again. It's the core for David Freeberg is herema Paly Hpeia and David Saacks calling in. I am your host, Jason Calans and we have a full docket today, I'd be ever remiss, though, If I didn't ask everybody for their fondest memories of the fifth annual all in sumummit produced by.Mr. David Freeedberg, David, you have any thoughts, How do you rate your performance as executive producer of the all in summit, You know what we didn't do what we forgot to do at the end of day, too, we didn't give a big shout out to our production team to Kimber to Lisa. to Nick. I think we Jason thought we should have brought him out. But I just want to give a big shout bring him out standing up. The team backstage absolutely crushed it, as always, this year. And then I had a great time. actually, You know, my favorite part.Is going to universal Studios. You just get to have a good time. It's great. Just. I'm reliving my childhood Shaath.😊With no trauma.By the way, shout out to our friends at De Longgi. I got one of those coffee machines that we had at the event.I put it in my kitchen.You know what I got. I don't know if you guys saw, but there was a water machine in the VIP area. I well. That is a zero plastic water machine. It goes through super filtration.No microplastics in the water. I have one now in our office. Thank you very much to them. I appreciate it.amazingaz company, Ss, you have a favorite moment from the show. Well, I think the highlight of the show, The thing that we'll remember for a long time was the president calling in during Jensen's talk.😊And I think Jensen's overall, the content and demeanor really helped calm down.This national panic.Over AI.It felt like we were the center of this national conversation. And I think it happened very.Organically, you remember, there has basically been the Cyop for the last few weeks leading into the summit to convince everyone that.Humanity was gonna to.Go extinct unless we did exactly what the Democrats want.And stop all AI development. And then, you know, we had first Sacha came out and.Kind of calmed everyone down and then Johnson did the same thing in the president.Called in and did what he did best, which is Cs.Total control of the narrative.That I'm sure these domer groups have been carefully planning for weeks and kind of thwared that whole operation.Nerurds enough. Get back to work. You have a super intelligent.Trump intelligence is now the term. We're going to call it Trump intelligence going forward. But I think you make a good point. A lot of people wanted to know.Treeberg, if this was planned.Was the President Trump calling.Flanned free part. It could take us behind the scenes.Basically, Jensen was connecting with him right backt and said, hey, call me back. And then he called him while he was on stage. Theres some texting going on.No, it was completely spontaneous. It was spontaneous. Yeah, It was not like a planned thing where Jensen was like, hey, let's do a whole thing. And so that was pretty amazing to have.😊This conversation that Jensen was waiting for from the president to happen live and.Clearly, this was the topic to Ju and DC as well as in the news and the media, and it was amazing to see it all happen on stage of us.And yeah, just to be clear, I thought.It was a bit. And I was like, well, put him on speaker phone. And Jensen says to me, how do he do that. And I just handed him. I always give an extra mic there in case somebody's mic goes out.P them on speakerphone, with the microphone.And President Trump, the funny'est president in history, funny is politician history.Did you notice.Ss that he drops a one liner through.A speaker phone perfectly.Perfectly, Jensen makes the greatest chips, everybody tries to reverse engineer them impossible, impossible, but it can't figure out how to press.The speakerphone button, okay.哎呀世先 do。😊How does he.Do it. It just drops the perfect line, but it.😊Tathin all seriousness.The nerds were losing their mind last week and the adults came, daddy came.Elong came.Jensen came, all the adults in the room came and they just said, listen.If you have software.That's defective.And not ready to put in the world.Maybe don't put it in the world and make it a little bit.You know, as you like to say, b brittle. So your thoughts on where we are.In this national discussion, which went from slowdown to maybe.Just release.Non bttle, resilient software.Well, I think that there was this.Classification that happened last week.Between.Up and coming organizations that.Want to call themselves labs.But are, in fact.Companies.That have P And Ls and shareholders and have to abide by the same rules that.Everybody else does.And then, the sophisticated companies.That have lived.In scrutiny.For many decades and have learned the hard way.The process of making things robust and the cost of making things not robust. So what do I mean,...(已截断)