May 30, 2026 ยท 2026 #19. Read the transcript grouped by speaker, inspect word-level timecodes, and optionally turn subtitles on for direct video playback
Edit labels for this show, save them in this browser, or download a JSON override for the production folder.
Transcript Playback
Human Agentcy
Human Transcript
Timed transcript
Blocks are grouped by speaker for readability. Expand a block to inspect word-level timing.
Andrew Keen
Hello everybody, it is Saturday, May the 30th, 2026, the last day in, or the last Saturday in May, and of course it's time for That Was The Week, our weekly summary of Big Tech News We Haven't Done It For 10 Days Last Time With Keith Teare We Asked Whether Capitalism Has A Future In The AI Age That Was Back On May 20th So We Haven't Done A Show For 10 Days Before We Went Live Keith Said To Me That It Wasn't A Big News Week This Week But I Wonder Whether He Was Conveniently Ignoring The Big News The Big News Of Course Of The Week Is That Anthropic Not His Team His Team Is OpenAI Has Become The World's Most Valuable AI Startup 265 Billion Dollars This Week Putting Its Value At 900 Billion Dollars Almost At A Trillion Dollars Way Ahead Of OpenAI's Last Valuation Of 730 Billion But As It Happens On Keith's Notes This Week He Talks About Something Called Agency Which We'll Come To There's No Mention Of This News So Keith Have You Conveniently Ignored The Big News Of The Week Or Do You Not Consider It Significant Will You Tell Me That It Doesn't Really Matter
Well, you heard it here first, but it's both true and BS at the same time. It's true that they did a funding round at that valuation and OpenAI is just slightly behind on funding rounds. That's true.
It Is What It Is It Looks For You On That Two Weeks So You're Saying That By Mid-June OpenAI Will Have Overtaken Anthropic Again Well OpenAI Is Pretty Far Ahead
On IPO Filing And We're You Know Just Behind SpaceX So I'd Strongly Expect OpenAI To Get Through The Gate First Now That Is Going To Shine A Light Because Both Anthropic And OpenAI Are Going To Have To Write S1s That's The Document You Write For The SEC That Investors Get To Read And The S1 Has To Have Audited Accounts In It The Most Recent Ones Can Be Unaudited But It Has To Be Accountancy Rubber Stamped Accounts And The Talk Of The Town Is That Anthropic's Revenues Are Basically Lies
They're misleading people. Om Malik, go to Om's website, anyone listening, and look at his most recent post, which is going to be in next week's newsletter because it only came out.
Yeah, and then go and listen to yesterday's All In Podcast discussion about this. It is pretty well received that Anthropics S1 will not have the revenue numbers that have been leaked because they're not real. Now, that doesn't mean the numbers are fake. It means they're counting things in revenue that they're really not allowed to count. And That Is And The Second Sub Theme There Is That Dario Amidai Is Deceitful And Devious Yeah You Don't Like Dario We've Had That Many Times It's Not Whether You Like It
It's Whether It's Just Factual Can't We Just Have Some Love For Open AI Can't We Credit Them That They They've Done Rather Well I Mean They're Getting Backing From Green Oaks Sequoia Altometer Dragoneer The Top VC Firm So Presumably They've Looked At Their Books They're Not Going To Pour Billions Of Dollars In Without Actually Knowing Exactly What They're Investing In I Mean You're Not Wrong That's A
that's a slightly different question than the one you asked, which is the competition with OpenAI. If you just asked me about Anthropic, they released Opus 4.8 two days ago. Honestly, they're fantastic. Their stuff is fantastic. Just to give you a personal anecdote, at SignalRank, we are moving to a regime where we have to have a third party value all of the holdings that we have in the index. So We Can't Do It Ourselves Anymore And Yesterday That Third Party Delivered To Us Their First Assessment Of The Value Of Each Of Our Assets And I Had An Agent That Was Creating A Valuation Report Monthly For Our Internal Valuation Committee And It Had To Completely Change Its Rules And Produce The Report Using This New Third Party Stuff Including Making Databases Analyzing And Then Blending And In About An Hour Clauded Completely Changed The Entire Agent Flow And Produced An Outstanding Report That Would Have Taken Me Days To Produce So It Is Definitely Fantastic So That So That's Not Really An Issue And I Think Dario By The Way As A Technical Leader Of A Team Is Also Fantastic I've always characterized him as slightly juvenile, making mistakes. I do think that's also true, but it doesn't take away from the other stuff. Right.
I mean, when it comes to slipperiness or being juvenile, I'm not quite sure how he compares with either Sam Altman or Elon Musk. That's another question.
Well, Altman's grown up. I mean, he really has. You could use some dated stuff, but if you look at his actual statements the last, I'd say, two months, Very Different Tone Like This Week Altman Talked A Lot About UBI And Jobs And Whereas Amadai Is Still Fear-Mongering About The Devastation That's Going To Be Reeked By AI Altman Is Saying UBI Is Not Going To Be Good Enough What We Need Is Ownership So Humans Have To Get Ownership In The AI Platforms In Order That The Wealth Growth Benefits Them And He's Changed The Narrative From UBI To Ownership And You Know That's An Interesting Conversation So I Don't Like The Fanboy Stuff And You're Right To Characterize Me As Tending More Towards Altman Than Dario On The Edge That I've Characterized Both Of Them As Not Good Leaders
And I think that But you have to admit Keith I mean I know it's easy for me to say this And hindsight's always 20-20 That a year or two ago You always used to say on this show Oh no one can catch OpenAI They're going to be a 5, 10, 15 trillion dollar company They've already won They've sewn up the revenue Everyone's using them They're the new Kleenex Or Google Or Whatever AI Brand And That's Wrong I Mean Anthropic Have Overtaken Them Well You're Using The Point On You Know Dario Is Certainly Not The Pope We'll Come To The Pope Later But They've Overtaken Them And Their Valuation Has Doubled What In The Last Six Months Whereas Open AI Seems A Little Stalled To
Say It's More Than Doubled Based On Leaked Revenue Numbers And I'm Predicting On This Show That Will Come Back To Bite Them And But It's Interesting Andrew Because You're Not One To Normally Take What VCs Do As Gospel In This Case You You're Swallowing Whole What The VCs Are Doing Are You Saying The VCs Are Being
Look, the VCs are at media. No, the VCs are acting rationally, but they're playing a different game. The game the VCs are playing is that they'll make two, maybe three times their money by investing these large sums this late prior to an IPO. And they probably will. But, you know, yesterday there was a guy, he said he was offered $10 million worth of Anthropic stock at a trillion dollar valuation. This Was Ohm By The Way. Ohm Was Offered 10 Million Dollars. It Was A Friend Of His. And They Turned It Down Because When They Look At The Numbers, There's No Justification. Now, That's True. But It's Also True That If You Invested A Trillion, You'll Probably Make Two To Three Times Your Money At An IPO. That's Also True. And That's Because We're At The Stage Of Everything Where
You Know The Only Way Is Up So But Looking Back I Take Your Point And We Will See In The Future I'm Not Gonna Call You On It But We Will Do Many Shows Many Shows On This Subject We Have Done Many And We Will Do Many In The Future But Looking Back And This Isn't So Much Of A Critical Question What Did You Get Wrong What Did You Misunderstand About Because You Were Convinced That OpenAI Had Sown It Up That There Was Going To
Be So I'll Ask Ask Answer A Question With A Question Why Do You Think I'm Wrong In My View Of The World OpenAI Is Executing Better Codex Is Unbelievable In Terms Of What It Can Do It massively dominates consumer use. Yeah, but not corporate worldwide. It is now catching up in corporate. It wasn't really playing in corporate. It's now catching up. And fast, I think. They just hired the head of Sales From An Enterprise Company I Forget Which One It Was They're Doubling Down On
Other Words Open AI I Don't Think I'm Wrong I Think I Think Open AI Is Building They're Both Going To Be Very Valuable Let's Be Honest But Open AI Isn't Losing Here It's
Friend Of Home If You're Listening Invest At The Trillion Dollar Level You Mention That Neither Sam Altman Or Dario Emadai Are The Pope And Of Course There's Only One Pope And Apparently I Know This Was Slightly Older News But We Haven't Done A Show For 10 Days The Pope Now Of All People Is Disrupting Silicon Valley And Interestingly Enough Is He's Doing It Not With Open AI Executives But With One Of The Co-Founders Of Anthropic
Christopher Ola I've never heard of him we always joke on this show about how many co-founders they seem to come out of the woodwork every time so what do you make of this Keith I know it's slightly older news and it may have come in your newsletter the week before but you included John Thornhill's excellent FT piece on hope disrupting Silicon Valley is this a big deal?
Well you can learn from everything so what do we learn from this? Firstly we learn that anthropic sent one of its founders to ask the Pope's help in protecting the world from AI. Well, what is that all about?
No, he did. The FT is only picking up on one side of it. And the Pope, to give him some credit, emphasized the importance of AI. Whilst Also You Know Not Liking Its Military Uses Which Of Course Is An Anthropic Mantra And So What This Really Is Is A Doubling Down By Anthropic On Trying To Create A Space A Moral Space Where It Is The Benign And Concerned I'd Say Benign And Concerned Are The Two Elements There Seeking From All People The Pope To Come On To Their Team Whereas Open AI Is You Know Benign And Unconcerned And Argues That They're In Control That The AI Maybe
The Open AI Should Should Send Their Co-Founder To The Chief Rabbi Or The Head Of The Rome Or The Head Of Another Church Or Maybe Even To The Head Of The Islamic Faith, but I think you're being a little unkind. I mean, you're like one of these parents who, for some reason or another, the child cannot satisfy them. Anything they do, you're critical of. You don't give open AI any credit for trying to work with the Pope. Anthropic, you mean. Sorry, Anthropic, to address this. I mean, there was an interesting piece I Know You Didn't Include It In The Newsletter From Noma It's A Very Good Online Magazine Yeah It Is About What They Call The Universal Destination Of Goods They've Been Calling For A Year Or Two About Policies That Should Induce As Many Many As Possible Of The People To Become Owners Of Capital You And I Have Talked About That I Know You're Sympathetic To That And That's Something Else That The Pope Addresses
In This Magnifica Actually The Only One Of The Two Embracing That Is Altman That Was Exactly Altman's Point This Week So Well Maybe He Should Have Sent Someone To The Vatican Or Maybe He Should Have Shown Up Himself Well Do You Really Think The Vatican Matters I Don't Well It Got A Lot Of Press Are
Well, except that it has a lot of political power and significance, that it's made significant news. And the fact, I mean, I agree with you in some ways, but on the other hand, the fact that the Pope of all people is getting involved in conversations about Silicon Valley, that the Pope in Thornhill's words is disrupting Silicon Valley, speaks of how this technology now, for better or worse, is so central to everything.
Well, what really happened this week is the opposite, is that Trump blocked An AI Bill That Came From Congress On The Grounds That He Doesn't Believe Government Stepping In To Have A Role In AI As A Arbiter Of What Can Be Released Makes Any Sense And Will Slow Down America's Development And So You Know The Real News Of The Week Is The Opposite Of What The Topic Was Asking For Has Happened Which Is AI Still Got Independent Authority To Determine Its Own Future
Yeah, no, Anthropik hasn't done badly out of that. Let's so we will see this is an ongoing conversation. But Keith Teare clearly does not like I'm not sure if it's Anthropik or it's just it's just Dario you don't seem to like but we will come back to that.
Well, Dary, if you're watching, even if you're almost at a trillion dollars, Keith Teare says you're making some mistakes. Maybe, will you give him some lessons, Keith? Private consulting, tell him how to run a company properly.
No I do it for free I enjoy it There you are Dario Keith Teare will give you some free advice on how to run your company Let's move on to your editorial this week about something called Human Agency but it's for people listening can't see the screen it's spelled A-G-E-N-T-C-Y it's not a typo what does this word agency Keith mean for you?
It's derived from a bunch of the articles that are in this week's newsletter that are talking about this step change in the use of AI. Thomas Tungers is One Of The People Writing About It But There's Several Reed Hoffman Talks About It
And Oh Malik Response To Yeah And The Tonga The Tonga's Piece Skill Distillation I Found Very Interesting The Malik Piece Is Called The Copy And The Guru And Reed Hoffman Features In A Wall Street Journal Piece About Digital Twins So It's An Important Issue This Week So What Are They All Arguing Or Agreeing Or Disagreeing On
They're Disagreeing Reid Hoffman I think is a little bit dated in describing AI as a digital twin and he's created a digital Reid Hoffman and I can't remember how you can go and use it, but you can go and ask him questions and it sits on top of all of his body of work. He's a busy guy, really.
He's a very nice fellow. I agree with you. Ohm reacts to that and saying, look, the purpose of AI isn't to be your digital twin. It's not to make many copies of you. It's Actually To Allow You To Carry Out Tasks That You Otherwise Wouldn't Be Able To Do Because It Would Involve Too Much Time And Effort And So You Can Multiply Your Outputs Using AI And That Is Agency Agent C
I thought of it myself. AI actually had a very boring view of what the Title Should Be This Week, and I said in one word, agency, and it said, oh, that's so much better than what I suggested.
I'm going to ask, I mean, after we finish this, I'm going to ask Anthropic if they think it's a bit cheesy. Anthropic wouldn't let, Claude would tell me that that was a bit cheesy, but maybe they're a bit higher end than Open AI.
Yeah, but you're right to ask. Originally, I didn't have the word human before it. I just said agency. And I Decided It's A Bit Too Subtle So I Put The Word Human So Very Subtle Agency And That There's Human Agency And The Point It's Making Is That These Agents Are Extending Your Ability To Get Stuff Done That They're Not Autonomous And They're Not Independent They're Really Your Your Your Your Assistants If You Will Or Better Than Assistants Because They Actually Do Stuff The Example I Gave Earlier About The Valuation Exercise At SignalRank Is A Great Example Of It And Suddenly If You Look At Your Week's Work What Do You Have To Get Done If You Understand How To Use Agents And By The Way You Don't Have To Install Anything Anymore Agents Are Now Built Into Both Claude Code And Codex The OpenAI Equivalent So you can tell them, create agents for each subtask here, and it will do it for you, and the task will get done much more efficiently.
So I said to you before we went live, I couldn't agree more with the own position or your position, but isn't it kind of obvious? And you said, no, a lot of people aren't using AI in that way. They're seeking digital twins. How are most people using AI?
It's interesting, Andrew, if you look at the user interface for the Claude app on a Mac, there's three tabs. The first tab is chat. The second tab is co-work. And the third tab is code. And chat is heavily dominant in use. People are just chatting. Looking For Answers To Questions Which Of Course Is A Good Use Case For AI And It's Better Since It Became Less Hallucinistic Cowork It's Where It Gets Access To Your Browser And Your Let's Say A Word Processor Or Your PowerPoint Or Your Excel Or Other Tools And Can Work Carry Out Work Using Your All The Tools On Your Computer That Was Pretty Powerful Not Many People Are Yet Doing That I Mean Obviously Quite A Few Are But It's Not The Dominant Use And The Third One Which Is Code Is What Developers Are Using It For And Some Non-Developers Who Are Learning To Code For The First Time While Not Learning To Code Learning How To Produce Code Using AI So I Think The Middle One Co-Work Actually Ends Up Being Super Important But Is The Least Used Of The Three
Yeah, you refer to the Tungus piece. He's one of your favorite writers, highly sophisticated guy in Silicon Valley. He Has A Piece Called Skill Distillation And I'm Quoting The Intro He Said I've Been Using State Of The Art Models To Teach Small Models Running On My Computer How I Work My Personal Agent Based On Pi Runs My Inbox My Deal Pipeline My Blog Publishing My Calendar And My Research It Looks Like A Chat Box It Looks Less Like A Chatbot More Like A Small Operating System So What I Found Interesting About This Is He's Clearly A Highly Sophisticated User Investor Knows His Tech But This Is The Future Ultimately Keith For All Of Us Isn't It I Think So
Basically You Get Surrounded By You Know A Paraphernalia Of Stuff That Makes You Super Efficient At Getting Stuff Done Some Of Us At Least I Mean The Tunguses Of The World Maybe The Keith Tears But Not Everyone But Certainly People Who Know How To Use It Yeah Not Everyone I Mean My You Know My Team At Signal Rank I Say Is Pretty Typical And They All Buy Into AI They All Use Well I'm Sure If They Didn't You'd Fire Them Probably Not But They All Use AI But They Look At My Use Slightly Weirdly I'd Say But The Results Are Getting Better And Better And Better And So There's There Used To Be A Lot Of Pushback On My Use And Now There Isn't Because The Outputs Justify The You Know Doing It So It I Think
It's A Process And So Where Will We Be I Mean Tungus I Mean I Said To You Before We Went Live It Seems To Me We're Still In The Pre Almost The Pre-Web Browser Moment In AI Where Are We Going To Be In Five Or Ten Years There's Going To Be Software Hardware Interfaces That Will Just Make This
That What We're Witnessing Is Human Evolution Not AI Evolution That AI Is Not A Separate Autonomous Scary Or Dangerous Thing It's Elevating Humans To A New Capability Level And Therefore This Whole Narrative As If AI Is Something Other Won't Be Able To Survive Because As People Use It It Becomes Clear That It's Your Thing Not A Thing
Yeah, although the Tunguses, as I said, of the world are going to be so far ahead. In five or ten years, they'll be on to something else. We will see. It certainly speaks of the concentration of power. I mean, the Tunguses of the world are already powerful and wealthy. They're going to become even more powerful and wealthy. Your wife, Jeunetir, talks about the concentration of venture capital this week. No surprise. What's happening on that front? What does Jeunet tell us, Keith?
Well, she's she's reporting the news and the news is that a vast proportion of all the dollars being invested is coming and this is true at every stage of investing from less and less funds into less and less companies. And that therefore, venture capital is not what it used to be. If you are a startup or a founder trying to attract capital At The Seed Stage Unless You Fall Into The Sequoia Capital Andreessen Horowitz General Catalyst Lightspeed Ventures Yeah The People Who Just Put All That All
Boxes It'S Going To Be A Struggle For You To To Raise Capital Now Last Week When You Were Away I Wrote The Editorial About Venture Capital And I Made The Point That Paradoxically This May Be A Good Thing For Who Venture Capitalist Or Entrepreneurs Actually For Pensioners What Was Big Picture Venture Capital Has Not Been Returning Money To The Source Which Is Pension Funds Since 2021
Investments Are Being Lost Large Large Amounts Of Money Being Lost And Funds Underperforming And So There Was A Liquidity Crisis In Venture Capital And Paradoxically This The Rise Of AI And The Concentration Is The Saving Grace Because If You Can Invest Hundreds Of Billions Of Dollars And In A Very Short Space Of Time Because Of IPOs Return Two To Three Times Your Investment That Money Will Flow Back To Pension Funds And Ultimately Will Provide Capital For The Early Stage Funds As Well As Everyone Else And So What We Might Be Witnessing Here Is A Fix Not Something Broken But Actually A Fix For The Entire Venture Ecosystem For That To Be True These IPOs Have To Happen Be Liquid And Money Has To Be Returned To The Pension Funds That Originally Gave It That
Anthropic Will Be Very Interesting But I Think Also Genet's Point This Has Been Happening For A While Now This Concentration Of Power And Wealth Of Course Venture Capital But It'S Everywhere Which Also Is Explains Why The Pope Has Become Involved In These Conversations About Figuring Out And I Know You You've Often Talked About This About We Got To Figure Out Ways Of Redistributing Both Power And Wealth So The Fact That The Pope Is Involving Himself In This Conversation Even If It's With The Help Of Anthropic Is Presumably A Good Thing
Transformation Hasn't Happened Yet It's Going To Take While I Mean There Is Some Skepticism And The Guardian Wonders Whether This Anthropic Alliance And The Little Skeptical Like You The Alliance With The Pope On AI Is A Form Of Vatican Washing
Which Is An Interesting Way Of Putting It Yeah I You Know Two Things Can Be True At The Same Time I Think Probably There's Truth In That If It Feels As If Anthropic Is Highly Strategic In What It Chooses To Do Based On Its Internal Needs Neither Anthropic Nor Open Eye Can Yet Be Trusted With A Societal
Level Narrative So Who Do We Trust Then And There Was A Piece In The Verge Which You Connected To An Interesting Interview With Sundar Pichai The CEO Of Google Can We Trust The Grown Ups I Mean Sergey And Larry Joked About Bringing A Grown Up In When They Brought In Eric Schmidt To Run The Company At The Beginning Of The Century When They Became A Real Company Is Google Maybe Amazon Maybe Microsoft Are They The Grown Ups Are They More Trustworthy On This Front
You know, I characterize Google as the most capitalist of them all. In a good or a bad way? Both. Look, here's the question that needs to be answered. How can the benefits of capitalism end up producing a better society?
Leveraging the growth of wealth to benefit society as a whole. Not through welfare but through uplifting everything. What is that? And You Know For That To Be Answered Altman Is Arguing For Ownership Mass Ownership Of These Companies I Don't Think That Really Cuts It I Think It Eventually It Comes Down To You Know An Impossible Problem Which Is Capturing The Wealth That Is Privately Owned For Common Good
Well, Elon honestly is closer to understanding this than anyone. He hasn't got a practical plan yet, but he at least can describe the processes that are going to happen.
His Concept Of The Lean Startup Which Acquired A Great Deal Of Popularity In The 20 Teens An Iconic Book Here's The Book Andrew You Got The Book You're Waving The Book Well He Has A New Book Out Called Incorruptible What Seems To Me What Was Interesting About My Conversation Is Clearly He's Being Pushed Around By The Zeitgeist He's Gone From Fetishizing The Lean Startup To Talking About How Companies Can Be More Mature And Responsible But I Thought That The Conversation With Rees I Mean He Doesn't Speak On Behalf Of Silicon Valley Lives In The North Bay But He Still Sees Morality Coming From Within Companies So His Model Of Companies Like Patagonia Moral Companies Maybe Anthropoc We Talked About That Do You Believe In That Rees Model Of The Only Way In Which We're Going To Get A More Moral Capitalism Is By Corporations Being More Moral Keith
No, I'm pretty happy with corporations just being purely profit driven. Of course, there's limits to what you can do. The law governs that in pursuing profit. I'm more interested in what happens to the profit. I don't really want to You Know It's Like Saying To A World Champion Boxer You Should Punch More Soft Because He Might Hurt The Other Person That Doesn't Make Any Sense You Don't Want A
Honest I Mean I Don't Look To Companies As Being You Throw Up Live For Us If Pushed Hard Enough Quite Possibly But, you know, I don't look to companies to be moral guides. I want them to be good companies. You know, when you invest in the stock market, you want them to be growing fast and making profit. That's it. There's nothing more to it.
I don't think it's the company's job. I do think Sundar personally should. I Think Individuals Who Care About The Future Absolutely Should Start Making Plans For Societal Good But The Job Of Their Company Is To Produce Wealth Okay So So
We've We've Established That So Where Does The Moral Guidance Come In Come Back Maybe We'll End Where We Began With The Pope Keith I Mean It's Got To Come From Somewhere
Honestly, you end up with You end up having to discuss politics Let's call it the In sociological terms The superstructure And civil society Our politicians are Massively disappointing Massively disappointing
Yeah. And in the absence of that, what you get is populism. Because the people care. The people care about the future. And if politicians can't be trusted, they end up, you know, going elsewhere. There's a couple of articles already in next week's newsletter about liberal, liberal nationalism, one from Noah Smith. And I Do Think That Populism Is The Unintended Consequence Of Lack Of Leadership
Was It A Big News Week Keith You Began By Saying It Wasn't I Think You Have To Acknowledge It Was A Little Bigger Than You Acknowledge I Yes I Was Wrong And Can We Credit Anthropic Is This Maybe You Should Slip One Of Those Pieces Into Your Newsletter To Acknowledge That The News That Anthropic I'm This Is My You Know Bad Year For Football So Anthropics My New Team I've Replaced Spurs With Anthropic We Have To At Least Acknowledge On This Show That This Is Big News Anthropic Tops Open AI At Least According To The New York Times Tops AI Tops Open AI To Become The World's Most Valuable AI Startup Keith Says That's Not Really True But It Kind Of True Yeah That's Big News
Look I think our listeners need to understand what's happening here is every week I write an editorial and a title and Andrew's job is to write an alternative editorial and title and what is just set up I guarantee this this is prediction that when you look at the keen on version of this show and compare the top bullet points it's going to be the opposite of mine
Fake News in other words well Keith as always you're a good sport excellent conversation we will talk again next week when no doubt we will tussle over the moral credibility of your friend Dario Emidai have a good week Keith and we'll talk in a week's time will do bye everyone