Did Anthropic Use the Pope as a Marketing Stunt? Ft. Amir Efrati

Summary

---

More or Less — Episode Summary

Episode Title: Did Anthropic Use the Pope as a Marketing Stunt? Ft. Amir Efrati
Original Air Date: May 29, 2026
Hosts: Dave Morin, Jessica Lessin, Brit Morin, Sam Lessin (away)
Special Guest: Amir Efrati (Co-Executive Editor, The Information)

---

Main Theme:

The Role of AI in Society, Regulation, and the Vatican’s AI Encyclical

This episode dives into the collision of AI progress, government regulation, and the recent Vatican encyclical on AI. The hosts and Amir Efrati debate the marketing impact of Anthropic’s presence at the Pope’s event, broader implications for AI adoption and regulation, societal resistance to technology, and the real-world limitations of current frontier AI models.

---

Key Discussion Points & Insights

AI Adoption — Flatlining or Just Early Days?

• Stagnant AI Adoption Curves:

  • Multiple hosts mention recent Federal Reserve data indicating a plateau in AI adoption (17:27, 00:10).
  • "The Fed introduced their latest data about AI adoption growth and it’s actually been flat or linear, nearly flat for the last three months." — Brit, 17:27
  • Venture and technical insiders debate causes: price, usability, and limited killer use cases.

• Price and ROI Barriers:

  • High costs for AI tokens and compute are cited as major adoption friction, pushing users toward open-source or less sophisticated models (17:44, 18:47, 20:15).

• Frontier Models — Amazing but Pricey:

  • Enterprise users see huge productivity gains, but the ROI (especially measured in “token productivity”) remains unclear (20:15-22:10).
  • Example: CTOs and CIOs (e.g., Uber) voice skepticism about incremental returns on AI spending.

---

Anthropic, the Vatican, and the Marketing Controversy

• The Event:

  • The Pope’s recent 43,000-word encyclical on AI drew global attention, with Anthropic’s co-founder among the Vatican attendees (7:08–9:21).
  • Hosts used AI to analyze the encyclical’s authorship, joking about the percentage “written by AI."

• Debate — Religious Moment or Marketing Stunt?

  • “I’m concerned by the way that Rome got taken advantage of by Anthropic for marketing. I think that’s actually the story here." — Dave, 8:32
  • Dave argues Anthropic’s Vatican appearance was a self-serving “marketing moment,” overshadowing the Church’s neutral message. Others see benefit in bringing technical leaders to the table.
  • Amir plays devil’s advocate, suggesting it makes sense for the Church to involve current AI leaders given the breadth and urgency of the issue.

---

AI Regulation, National Security, and Geopolitics

• Executive Order Limbo & D.C. Drama:

  • Update on the Biden/Trump administration’s pending executive order on AI—reportedly stalled and under industry/government negotiation (5:16-7:00).

• National Security & Model Access:

  • Anthropic’s relationship with the DoD: balancing model guardrails vs. national security needs.
  • Amir: “Much of the national security establishment relies on and loves Anthropic models and they want to keep using them...” (5:43)

• Foreign Interference & Data Center Activism:

  • Dave draws parallels between student activism against data centers and previous opposition to TikTok, raising questions about foreign (Chinese) influence (13:05–15:08).
  • “I actually started asking around, is this China? ... And then I saw over the weekend that Mr. Wonderful from Shark Tank ... hired forensic experts to go and factually prove that it is China actually funding all of this opposition in Utah.” — Dave, 13:50

---

Bottleneck: Cost, Infrastructure, and Scaling

• Anthropic’s Scaling Issues:

  • Demand is outpacing compute supply, resulting in high “luxury” tokens prices and frequent downtime; foreign-funded activism allegedly threatens new data center capacity (15:32–16:14).

• Government Access to Models:

  • Speculation around government getting privileged early looks at “frontier” models (like Anthropic’s Mythos) for oversight and security (9:51–10:52).

---

Usability, “Normies,” and the Future of AI Agents

• Adoption Beyond Techies:

  • Extensive back-and-forth about whether average people will ever want or need personal AI agents (24:10–34:38).
  • Jessica argues most people won’t pay for bots—"even if their lives would be so much better."
  • Brit counters: “Moms will always pay $20 a month if like literally 2% of their mental load gets shaved off.” (27:22)

• Key Quote:

  • “This is a fundamental mistake I think the technology industry makes: every human on the planet wants to march toward efficiency. And I think just because a better product exists, even if it was free, some people just wouldn’t adopt it.” — Jessica, 28:40
  • Dave responds, “The great thing about AI is that we can now just talk to computers with English...the thing we’re trying to get to is that computers are no longer suck in our lives...” (34:38)

• Spectrum of Use Cases:

  • Amir proposes the market is at “1% penetration,” and predicts mass adoption after significant improvement in user experience and discovery (36:04–37:22).
  • Personal anecdotes from the hosts about both mind-blowing use and ongoing skepticism in everyday life abound.

---

Socio-Cultural Topics and Pop Culture

• Offline Experiences and Digital Detox:

  • Dave shares a story about a “dad's trip” off the grid in Zion, reflecting on the value of disconnecting from the internet (41:54–43:37).

• AI in Entertainment:

  • Brit highlights rapid developments:
    • Suno’s $5B raise, 11 Labs’ music genre fluidity, the resurrection of Stan Lee’s voice for Marvel (43:37–45:21).
    • Spielberg’s warning at Cannes about a lack of original stories.
    • “AI Canva” moments emerging—tools enabling anyone to generate high-quality content.

• Book & Media Recommendations:

  • Jessica recommends Remarkably Bright Creatures and A24’s Eternity (45:21).
  • She also shares a story about visiting Crunch Labs, meeting Mark Rober, and hitting the bullseye on their tennis ball cannon (45:25–48:54).

---

Notable Quotes & Memorable Moments

• "Is the word of God supposed to be written by AI, or is it all just the word of God?" — Dave, 07:34

• "I really think that the Church was taken advantage of and it became a marketing moment for Anthropic, not an encyclical moment for the Church." — Dave, 08:52

• “You’re actually dealing with foreign adversary that’s trying to divide us on this technology as well.” — Dave, 14:29

• "The productivity per token is like totally unclear in how to measure that." — Dave, 21:26

• "I am starting to feel ... the number of use cases on a given day is somewhat bounded." — Jessica, 22:14

• "It is this real experience ... I live in a world where most people don’t need ... a personal agent that can do all these amazing things." — Jessica, 23:28

• "The only way that you’re going to fit into that is by actually doing the hard work of user experience ... this is actually the number one thing that’s driving me absolutely crazy in AI right now is that nobody uses this stuff." — Dave, 33:44

---

Important Timestamps

• AI Regulation & Vatican Event: 05:16–09:21
• The Pope’s AI Encyclical, Authorship Debate: 07:11–08:52
• Anthropic’s Marketing Motives Debated: 08:32–09:21
• Foreign Influence & Data Center Activism: 13:05–15:08
• AI Adoption Plateaus & Price Friction: 17:27–18:47
• Frontier AI Models in Enterprise: 20:15–22:10
• “Normies” & Real Barriers to Bot Adoption: 24:10–34:38
• UX and Product vs. Consumer Reality: 33:44–34:38
• Offline Healing & Pop Culture Break: 41:54–43:37
• AI in Entertainment & Media: 43:37–45:21
• Book/Media Recs, Mark Rober Story: 45:21–48:54

---

Tone and Style

• Conversational, Collegial, Sometimes Combative:
  The hosts are insiders and close friends, so the conversation swings between playful jabs, in-depth debate, and personal stories.
• Cynicism vs. Optimism:
  The group oscillates between healthy skepticism about Silicon Valley narratives and genuine enthusiasm for future potential.

---

For New Listeners

• This episode pulls back the curtain on how the tech elite and journalism insiders view the AI inflection point: big on ambition, wary of hype, and deeply divided on both regulatory context and mass adoption.
• The Vatican controversy is both a window onto tech PR games and a symbol for how emerging technologies are vying for moral and narrative power in society.
• The biggest takeaways: AI isn’t as widely adopted as headlines suggest; technology always lags mass behavior; cultural and regulatory forces matter as much as raw tech progress.

---

End of Summary

Transcript

[A]

I think we're all going to start talking to ourselves in our cars, at our desks. We're just going to be talking to our AIs.

[B]

And if we see power issues come up, then Trump will say, well, you know, you need to pay for it.

[A]

Did you guys see that? The Fed introduced their latest data about AI adoption growth and it's actually been flat or linear, nearly flat for the last three months.

[C]

There were some charts running around yesterday that Claude Code and Codex have, like, flatlined, you know, their daily usage.

[A]

Eleven Labs is also resurrecting Stan Lee's voice and a partnership with Marvel, CTO

[D]

of Uber basically came out and said it. I mean, this had a huge impact. Saying, like, he's not seeing the ROI to spend more on AI.

[C]

This is a classic venture capitalist predicting the future, but not technically understanding where it was going to go at all.

[D]

More or less easy, no. Or is it?

[A]

Yes.

[D]

We'll debate the tech that's best when we get More or less.

[A]

Dave and Grit, plus Ham and Jess

[D]

put it all right to the test. More or less.

[A]

Why?

[D]

Hello, friends. Welcome to More or Less. And we have a special edition today. Mr. Co Executive Editor of the Information, Amir Efradi. Amir, welcome to your first More or less, I believe.

[C]

Amir, what are you doing slumming it with us, man?

[A]

Yeah, we are not organized like you are.

[B]

I'm a religious listener, and not just because my boss is part of this podcast.

[C]

Oh, interesting.

[D]

It's a nice.

[B]

I would listen to it anyway. So.

[C]

So who's your favorite? Yeah, just tell us.

[B]

Difficult, difficult to answer. But I. I will give a little shout out to Britt because I believe it was a week. A week or two ago, she highlighted the trend of developers whispering to their AI. So if you go into an office of a startup, you shall see many a whispering. And we wrote about that this morning in AI Agenda, which.

[C]

Oh, amazing.

[A]

Wait, wait, wait.

[C]

So wait, was she right or was she wrong?

[A]

First of all, I was right. Second of all, I was not just right two weeks ago, Amir. I was right a year ago. This was my prediction for the year. I might even be in 2024. Jess, maybe we should go back and check. In December, when I was like, I think we're all going to start talking to ourselves in our cars, at our desks. We're just going to be talking to them.

[C]

I'm sorry, but you did not predict whisper flow in the way that people are talking to them.

[D]

Exactly. We're going to have to go back. You predicted we'd Be talking to. Like, what were those things called? Echo devices.

[A]

I just thought we'd be talking out loud to our AIs a lot and like I would call whisper an AI that we talk to.

[C]

This is a classic venture capitalist predicting the future, but not technically understanding where it was going to go at all.

[A]

Oh my God. That was a dig, Dave.

[D]

Ouch.

[B]

But the prediction of the future is fundamental to what we are as humans

[C]

and how we communicate.

[D]

Well, guys, as you can tell, Sam is not here. So Amir is. As much as I think Brit, Dave and I bring so, so much to this pod. I was like, you know who we need? We need Amir because Amir, we think we're in the center of AI. Amir is in the center of the AI infrastructure world. The AI, these models. Like he has been leading this coverage for the information for, I don't know, 12 years. Has AI been around for 12 years?

[A]

I mean, depends who you ask, but yeah, but I do think I've always thought that journalists would make for the best investors. And we've actually seen many spin out to become investors like M.G. siegler and others. So I'm saying I'm validating Amir here because I think he's going to give way better insight than, I mean, you

[C]

want to join our firm?

[B]

Well, I think about Jack, you cannot.

[D]

What are you doing?

[A]

I don't have a non solicitation with you, Jess.

[C]

In this podcast before we started, he said that I was his favorite. But now that we're, you know, on air, he chose Brit. So now I'm going to recruit him to Offline Ventures.

[D]

He's just being political because honestly, he walks into the office and he's always like, know, you know, Dave made some good points on the pod today. Dave's on fire on the pod today, you know, so I, I know that you are presumably. He never brings up Sam.

[C]

He's had enough of listening to him around the. Around the information shop, I'm sure.

[B]

Well, I don't, I don't have. I can't compete on BMI with Sam.

[C]

None of us can. I don't use to take a lot more peptides, man.

[B]

I don't take peptides. I don't use Instagram. But. But I did just buy two gadgets for my kids that I could talk about if we go in that direction later.

[C]

So should we talk about the real news?

[D]

Let's talk about the real news. So let's pick up. And again, we had hope everyone had an awesome Memorial Day weekend. We can get into the some of the fun things in the latter half of the pod, I'm sure everyone was up to. But once again, AI occupying primo position. And let's start where we ended. So, Dave, I introduced, I talked about an article that recently published in the Information about briefing tech companies about this forthcoming executive AI Order. You, before I finished my sentence, called it fake news. And the news wasn't fake in that this meeting actually did happen. But you were right that this executive order.

[C]

You can say that again.

[D]

Pretty. You were right. I have no problem.

[A]

She's married to Sam.

[C]

No, no, I'm just modeling it for Brett.

[D]

Yeah, it appears that this thing is very, very much in limbo after some pullback from Trump. But, Amir, I know you're actually following where this is at this moment. So what are you seeing and what do you think is going to happen here? And then, Dave, we'll let you talk a bit about why you predicted this.

[B]

Well, the different factions are really interesting because while you have the fight between anthropic and DoD over these various guardrails that Anthropic was insisting on, it's still case that much of the national security establishment relies on and loves anthropic models and they want to keep using them and they want the relationship to be good or at least to have unfettered access to whatever the models are. Mythos. And the introduction of Mythos only kind of accelerated or emphasized that. But it's always been the case. And I think people at Anthropic and OpenAI and other labs, they understand, I think, fundamentally that, like, at the end of the day, the government is going to be involved in one way or the other with these models. So, yes, there's this EO and limbo. Maybe not a lot is going to happen in the near term, but through a combination of this becomes a national security kind of thing and the growing kind of backlash that you do see in public with kind of trepidation around job loss. You saw it in Rome over the weekend with Pope Leo, who got a tremendous amount of attention for his encyclical on AI. I don't know how many of you read all 43,000 words. I only read maybe 5,000 words.

[D]

I read the chat agency summary.

[A]

Why didn't your agency summary?

[C]

So I had my agent actually use a skill that I built to figure out what percentage of the encyclical was written by AI.

[A]

Guess what the answer was. Guess what percent was written by AI. 64 from the Pope. You think 64, Jess? Is that like a biblical number? Amir what do you think?

[B]

You know, I would guess a third.

[C]

10%.

[B]

10. 10%.

[C]

Okay, but still, even so, like, is the word of God supposed to be written by AI, or is it all just the word of God?

[A]

Well, this is God. It's AGI, Dave.

[B]

AI can be.

[D]

Is that within the margin of error of your bot?

[C]

It might be.

[D]

I mean, because 10%.

[C]

Well, these. These predictors actually predict the likelihood that it's written by AI, so it's a 10% likelihood that there are parts of it written by AI.

[D]

Okay, that's.

[C]

So it's relatively low.

[D]

Okay, so, Amir, you were saying? So do you think this palpal. It wasn't a decree, it was a. It was a word that I hadn't even heard before.

[B]

Yeah.

[D]

Cycle.

[B]

And then encyclical. Apparently they'd been dieting about this for a while, and a lot of people were anticipating it. First time I ever heard of it. And it was kind of symbolically important that an anthropic co founder was in the room for this in Rome with the Holy See. And so I think a lot of the.

[C]

I'm concerned by the way that the. The. That Rome got taken advantage of by Anthropic for marketing. I think. I think that's like, actually the story here.

[B]

Say more, Dave.

[C]

No, I mean, I think that's like such a journalist.

[A]

Such a journalist response.

[C]

Yeah, it's like actually the story.

[D]

I got to sit back, guys. My work on this podcast is. Is done.

[C]

It's actually the story because the. The Pope shouldn't have had anybody on stage with him, let alone, you know, one of many of these different researchers, providers. And so I really think that the Church was taken advantage of and it became a marketing moment for Anthropic, not an encyclical moment for the Church. I think that's pretty unfortunate because it could have been a. A much more egalitarian, neutral message that could have gotten people to think about these questions. Instead, it became marketing for Anthropic.

[B]

I think it makes sense from the Pope's perspective, because at least you're saying, well, look, a legitimate leader in the field is willing to come here and participate in this articulation of a broad set of concerns about AI causing a lot of pain while it also causes a lot of prosperity. So I could see why he did that.

[D]

Okay, so what's the net. Net, or where you think this regulatory action may be going?

[B]

Broadly, where we seem to be going, is the government eventually, one way or the other, having much deeper participation in the. I Don't want to call it vetting, but certainly getting an early access, early look into what is being developed. And that's just, isn't that the actual

[C]

real thing here, which is that these are powerful tools that now have national security implications in that, you know, these more powerful models are, they're like extremely powerful. They can, we're seeing it. They can find tons of bugs. I'm hearing that they can reverse engineer any software binary. And so like software is no longer closed source. All software will be open source going forward because you can just rip open any binary and reconstruct the software. And so like that means that there's like a massive national security implication to models that are more powerful than what we are currently using at the consumer level. And so it makes total sense that they would want to see them first. Right, and so that makes sense to me.

[B]

Dave, you're saying that closed source models are not closed and anyone can re engineer them at this point, Is that what you're saying?

[C]

Models, binaries. So like every app on your computer that is compiled into a binary that may not matter anymore, like it used to be that you couldn't open them up and figure out how they were written. Right. And so it appears as though traditional software may no longer be closed source anymore, which I think is a really interesting thing to think about.

[A]

Who are the people reviewing these binaries and like what is their background like within the White House?

[C]

What do you mean reviewing the binaries? Like, we're, we're confl. We're, we're conflating words. You mean models?

[A]

Yeah, the models. Sorry, who at the White House is reviewing.

[D]

I mean it's usually out of the cyber, homeland security, security kind of folks.

[C]

Well, and you have, you know, Sriram, who works in the White House from Silicon Valley that runs AI for the administration.

[A]

So we feel, we feel good about. Their reviews are coming out of Sriram's amazing.

[D]

I, I feel good. I also think, I mean the, so the question here, big picture again, very hands off approach to AI regulation from this administration and all regulation, but the drum beat of anti AI sentiment, pressure around jobs from voters getting louder and louder, administration facing more pressure to do something. You have mythos coming out where I think like DOD did get an early look and that's like actually held up as like a somewhat productive kind of relationship on both sides. But it was like a very narrow window. I think they had like a week or two ahead of time. And so those are some of the things being ironed out. But I guess we're basically back to where we were, which is very little drumbeat of meaningful regulation, which I just don't think will hold. Like the question this raises for me is like, what else will the administration do to appear tough on tech? Because they're going to have to. And the election is getting closer and closer and Eric Schmidt is getting booed at graduation speeches and well, so this

[C]

is actually something else I was going to bring up. You bring up the booing at graduation speeches. When I, when I initially saw that, I was asking the question like, why is this happening? This is much worse than it appears. We've talked about this on this pod. And then I had this like weird feeling that the pattern seemed very similar to something else we've talked about on the pod quite a bit last year, which is the TikTok ban. And that there's like a similar feeling to what's going on with this data center activism. Students at graduations booing. I mean, this is the same thing that was going on with Palestine. This is the same thing that has gone on with many different topics over the last many years.

[D]

And it's like, are you implying there's a common force behind them to destabilize the United States?

[C]

Indeed I am. And so I started asking around, I was like, is this China? To a lot of friends that you know, and folks that I know around Washington and various places. And then I saw over the weekend that Mr. Wonderful from Shark Tank who's been trying to build a data center in Utah actually started asking the same question. And then he hired forensic experts to go and factually prove that it is China actually funding all of this opposition in Utah. And he actually mapped it all out. You can go look on his Twitter and Instagram and he shares all of the data and apparently has now shared this with the national security establishment. And so I actually hadn't been thinking about it, maybe to my, I don't know, strategic detriment or something, but that we're actually up against much more formidable foe than just. And you know, anthropic is used in a fear based message to scare the American public. They're all going to lose their jobs and you know, et cetera. You're actually dealing with foreign adversary that's trying to divide us on this technology as well. Which, you know, makes sense of course, I guess that makes sense that that's happening, but I think it's kind of more in the water than people realize. And it's now showing up at city council meetings in like rural Utah and that's like a, that's a problem.

[D]

Well, we talked at length about what the tech community and tech companies can do to try to combat this and what bigger problem it is. I don't think we have to fully

[C]

open the Ban TikTok chapter because here it is again like causing social problems in the United States.

[D]

I mean guys, like we went through that and TikTok is still here and it's in this pseudo JV and does anyone really care? I mean it just sort of the difference here is.

[C]

But it's actually damaging our ability to innovate on a major, a really important technology tool. Right? And so that's a problem like Anthropic is also up against scaling problems where they are unable to scale their downtime is really, really bad. You can just go look at their website and look, they've got demand through the roof. Product market fit through the roof. Right. And so the prices they have to charge are unbelievably high. They're like ridiculous luxury prices for their tokens. And it's all because they need more capacity. Right. And so if foreign adversaries are successful in stopping us from being able to build the capacity for frontier models, then that's like a real national security issue. So I guess I get it.

[B]

Well, the Trump administration is unlikely to do something so heavy handed that it takes the economy. This is why going after Anthropic was quite, in calling them a supply chain risk was quite an interesting thing to do because if Anthropic were to fail, that would have a massive cascading effect on the entire economy. And AI and data centers in general are what is literally keeping this economy going. So I think they're going to have to find a very specific balance. Maybe they can go after these companies on speech issues and, you know, national security access. We have the whole scenario now where there is a model that exists called Mythos. Nobody really has it except for Anthropic. And who knows if the actual real model, as opposed to distilled versions of it that they give access to, will ever kind of come out. And maybe that's the thing that the government gets exclusive access to. But as far as the data center story, which is a very local story, I can't see the government federally doing anything, you know, to make life harder. And certainly all of the companies have made very specific pledges around power. And if we see power issues come up, then Trump will say, well, you know, you need to pay for it. So yeah, I think that's what's going to happen.

[A]

By the way, did you guys see that the Fed introduced their latest data about AI adoption growth and it's actually been flat or linear, nearly flat for the last three months. So it's interesting.

[C]

Yeah, there's a bunch of, there's a bunch of data that suggests this actually,

[A]

to see the curve starting to flatten out and like, where do we go from here? At least in the U.S. right?

[C]

I think it's just price, personally.

[D]

This is a great next topic.

[C]

Yeah, go ahead.

[D]

Okay. So Dave, you think it's price?

[C]

Yeah, I mean, I think they hit it. I think that you're seeing signals of this all over the place. Although I don't. It's hard to get really. Maybe you guys have better data. There were some charts running around yesterday that Claude code and codecs have like flatlined, you know, their daily usage. I asked around. I don't think that's actually true. But you know, there are a bunch of signals that suggest that there's kind of a flattening happening right now on a bunch of different metrics. And to me the only explanation has to be price. Right. Like it's just too expensive to use the frontier models. And we're seeing this in the open clock community where like people just like, they hit a ceiling really fast and they're like, I'm just not paying for this. And we're also seeing founders that are running massive agentic stacks and they're swapping out for open source models like left, right and sideways because it's just like unsustainable for their margins. Like they just cannot pay these luxury prices.

[D]

Amir, someone on your team. Sorry, but just relative to that, Laura on Amir's team, Laura Bratton, who writes our Applied AI newsletter, spoke to the CEO of Uber right on this point.

[B]

CTO getting this wrong.

[D]

CTO of Uber, where he raised the point. He just basically came out and said it. I mean this was, had a huge impact, you know, got a lot of attention. Saying like, he's not seeing the ROI to spend more on AI. And we've, we've done a number of stories. Nikesh Arora, Palo Alto probably told us this about six months ago, 12 months ago, when Deepsea came out, he said he was sort of keeping his budget relatively flat, but just getting more for it. Now that could have changed dramatically. Am I remembering those facts correctly?

[B]

Almost. So the CTO almost, almost bragged about blowing through the AI budget in the first few months of the year. Then the COO over the weekend in a podcast said that I think that this led to a bunch of internal meetings and that he. This is. Andrew McDonald did not see much of a return on this investment yet. But the thing is, like, it's way easier to spend on these models, Dave, than to like hire more people. And it's something that you can shut off unless you're using it in production very specifically. So I'm just curious how you reconcile what you were saying about the cost and everything with just the unprecedented growth that Anthropic has seen and the flippening with OpenAI where it has just completely overtaken OpenAI from a revenue perspective.

[C]

I think both things can be true. Right? Like these amazing frontier models like 5.5 and 4.7 are amazing and they are genuinely incredible at doing a lot of engineering tasks with code in particular. And they are also very good at coordination and project management with agentic harnesses. Right. And so then the other thing that's happening, we see this at openclaw, is that when you deploy agents into an enterprise, your token consumption goes up sometimes on the order of 3,000 per percent. Right. And so you go from having single individuals hitting these AIs 10 times a day to get information to having agents hitting the API 100 plus times per day just in like default mode, let alone doing work on behalf of people. Right. And so it makes total sense that you see this parabolic curve happening when once the agents came into the game because their consumption is just massively larger than humans. But at the same time, I think the article you guys did with Uber, that is reflective of startup CEOs that we're hearing from. It's reflective of CIOs that I'm talking to. You know, I spoke at Dell World last week and there were 10,000 people in the audience. I was talking to CIOs all day long and everybody's asking the same question, which is, what is the productivity of the tokens? And yes, it's amazing that we can do work now either by individual humans using AI to do work, or individual humans building agents that then do work. And yes, it's easier to build an agent than it is to go do a recruiting loop, but the productivity per token is like totally unclear in how to measure that. And so that's kind of the question everybody's asking. So that's why I'm saying both of these things can be true, that they're.

[D]

Here's something I've been wondering, and I'm just channeling Sam a little bit, because

[C]

what would be, please do more or

[D]

Less pod without a sort of wild and controversial but not entirely wrong point of view. I am starting to feel having like hooked up cloud code to various things as it relates to my own personal productivity, not like the productivity of everything across the information or whatever. My own personal productivity. The number of use cases on a given day is somewhat bounded. Right. Can help me with emails, it can help me with scheduling, help me with research. Right. And I can get a lot of value from it. But like, I know, Brett, you're shaking your head, but I. But the people are more like me than you. So let's.

[A]

Right now they're gonna be me soon.

[D]

I'm not sure they are. I. I think I. And I'm not an AI.

[C]

I know they are there.

[A]

I'm not for sure gonna be me.

[D]

Great.

[B]

I agree, guys.

[D]

I haven't even finished.

[C]

That's like being in like 84 and saying people aren't going to use computers. Like, it's just like, yeah, all right, I'll stay. I'll stay out of this.

[D]

This is not like an anti AI diffusion statement. Right. I believe AI is going to transform society. I think the ways I do, I really do. And, and I think the ways that I'm not letting you guys bully me into a different position, but I think

[C]

someone's got to play Sam, where a

[D]

lot of new is going to have to be built pretty fast in a way that normal people to. Can use it to realize a lot of these levels and increasingly like what shifted for me. So I think I totally understand that adoption takes time. Look at the history of the personal computer, look at the history of the Internet. We're in the pre DARPA Net days. I get all that. But it is also true that work is very defined and segmented by what people do. And I don't think AI is going to be as transformative to all types of work universally. And I think what we're likely to see is that there are some segments of the working population that are going to be able to deploy it to a much higher ROI and to a future of much more adoption. And there might be some that don't. And I actually think that it's very possible. We live in a world where over the next 10 years, most people don't need to be paying $20 a month for a personal agent that can do all these amazing things. And even if, you know, in theory, their lives would be so much better if they had this personalized health concierge or personalized this or that, they're going to Be like I don't really need it. Not because they hate AI but or they hate big tech companies but just like the universe of things that they spend their time on in a given day, like they're making a very rational trade off. Like and I think AI will transform work, but it's not going to transform all things equally. And I think it is this real experience. And I've talked to a lot of people, I mean you guys are as bot pilled as anyone on the planet. I'm not saying that is a bad.

[C]

Why do you say that? I don't know why you would say that.

[D]

Who are more bot pilled, very self

[A]

aware of it and they're kind of

[D]

like I've sort of built everything I need on a daily basis. Like I actually, you know, spend about x percent of my life doing this and it's gotten y percent better. So I think when we're thinking about what AI adoption curves are going to be looking like, I do believe economics is probably the best way to understand them. And you're right Dave, that price is probably the best way to understand a flattening. But I think as we continue we can't actually just acknowledge that the utility isn't going to be as transformative or as ROI positive universally. And I think that's going to look wildly different. And I think there are going to be segments of the population where it's totally rational for them to not have a bot be the center of their lives. I really believe that.

[A]

Okay, here's my take. I hear what you're saying. I have a lot of normie friends too. I'm from Texas and I think two things. One, like there are so many ways I use AI and have been using agents outside of normal chatbot like you know, prompting that normal people have no idea about and I think will want to adopt. Like I have created financial dashboards for our family. I understand like down to the penny.

[D]

Most people don't manage their finances as actively.

[A]

The number of people that like look at their bank accounts and just want a bank account update of like, like literally money is the number one thing people care about.

[C]

ChatGPT just launched a literal mode flat integration.

[A]

Yeah, with ChatGPT I think there's a

[D]

difference between like I think there are many people. That's a good analogy today, right? You can use like plain old Gmail or you could use Superhuman or you could use all these productivity enhancing things

[A]

but like you have to have the context on the person is the thing. And like, like I took my daughter to Legoland this past weekend. It knew things about her already. From all of the conversations and context that I have with my look, I

[D]

very candidly think most people don't care. I really do.

[A]

Okay, do moms and parents not care about summer camp signups, about all the school emails and the things they need to do? Like, moms will always pay $20 a month if. If like literally 2% of their mental load gets shaved off. Then the health thing you talked about, then finances, like, and then. Wait, wait, wait. I'm not done. I have two more points. One is, I think that the communication layer that has not been invented yet by agentic AI and the ability for, like, my agent to talk to your agent, to like, not bother you. But I just need a quick answer on something. Maybe it's about scheduling the next more or less or something else. Like, there's so much interesting stuff to think about there. When my son's teacher can, like, have a full personality profile on him based on what my agent knows about what's been happening with him behind the scenes over the last month. So she knows why he's being quiet in school and she can, like, treat him differently at school. Like, that stuff is going to be transformative. And the last thing I'll say is, I, I do agree with you that this is like a lot of friction for a lot of people and they don't want to spend their day just like with an AI agent and app. I think a lot of it's going to be invisible. Like you're going to be adding things to your shopping cart when you're shopping and you're not going to realize AI is running the background, figuring out the next thing to show you to buy. It's like TikTok is invisible with the AI neta has invisible.

[D]

But I'm not talking about that. Like, I just think history is full of example. Like, this is a. I'm pressing this because this is a fundamental mistake I think the technology industry makes that every human on the planet wants to march towards efficiency. And I think just because a better product exists, even if it was free, some people just wouldn't adopt it. I don't think that building the best free product guarantees adoption.

[C]

Like, I want to respond to this, but I also want to hear Amir because he's been making faces that suggest

[D]

great response in so many ways. Look, I think that's the fundamental thing. Like, and tech people assume that it's more efficient and cost effective. People will just adopt it. And it turns out people make A variety of choices that can range from very personalized choices for a whole lot of reasons. And there is a general march of fundamentally useful tools and technologies having wider penetration in society through different channels. But like there, I get, as I have three kids, I get as many school communications as the rest of them. I could make a million choices based on existing technology about how to be better on top of it, and I choose not to. And I just don't like, I'm not optimizing towards that. So I think we have to be very careful to assume that just because something is technologically possible and getting easier and easier that people will naturally adopt it. Like, humans make a bunch of different choices for a whole lot of reasons. And I think this is a very core issue. And you know, when I hear OpenAI say something like, who wouldn't want this amazing doctor in their pocket for all of this kind of stuff? I think some people want that. And some people would say, you know what, I actually still prefer to talk to a human. Or who wouldn't want a pernuvo body scan of your. Like, there are just different types of people who want different things. And the AI industry needs to like, learn this fast, in my opinion.

[C]

Look, I agree with you. There's always been this problem in Silicon Valley which is that the narrative gets way out ahead of the actual, like, user experiences, the customer experiences. Like the obsession with the technology goes like a million years ahead of where, or 10 years ahead, usually of where reality actually is. And the difference between today and a world where more people are actually using these tools to not just make their life more efficient, but perhaps have more fun, more delight in their life is usually user experience. And Apple, one of the number one companies on the planet, is proof of this, right? Like, the way that you create great technology experiences in people's lives is by actually spending the long haul on creating simple, usable, delightful technology that actually can fit into people's very busy lives. That you're absolutely right about. Like, people don't care about technology. They care about their family. They care about their emotional state every day. They care about, you know, making money to put food on the table. They care about going on some travel. Like, they care about, like, things we all universally care about. And the only way that you're going to fit into that is by actually doing the hard work of user experience. This is actually the number one thing that's driving me absolutely crazy in AI right now is that nobody uses this stuff. Nobody uses, you know, Claude Code and Codex in the grand scheme. Of the Internet. There's like single digit, millions people using it. They're incredibly technical. They're like computers from 1984 that you have to like, literally compile the kernel and like, talk to it in like this weird language. Like, it's still super, super hard to use. And those of us that have spent the time to like, tinker with it and figure it out, like, we're all like, this is awesome. Like, it's changing the way that I work. And actually maybe work is going to be completely different years from now. I fully understand this. It's like I've actually spent most of my career working in user experience. Like, the thing I am good at is user experience and design and product. And like, that's actually why I'm passionate about it. Because I want to go spend the next five to 10 years of my life working on agents. Because I do think that just like computers, somebody had to go on that mission and try to figure out how to make it usable for everyone. But, like, the broad conversation in Silicon Valley is like, because there's this like, new narrative game. And I think it's partially that, like, the new generation grew up with the Internet. They like, massively amplify each other into these, like, crazy narrative spirals. You kind of have this like, narrative gloss over the whole thing right now. And everybody's just trying to like, out narrative each other about their, like, new way, their new take on the current version of the technology. When, like, guys, we really need to go on a serious user experience odyssey to figure out how to get this stuff to more people.

[D]

Like, I just also think, and maybe this is putting a finer point on it. So thank you guys for indulging me in this. But, like, not everyone wants to. Productivity max.

[C]

I'm not talking about productivity, Jess. I'm actually just talking about delight in your life. Like, it's like most computers suck to use.

[D]

Not just you access to change. We need entire new use cases. Because right now these things are geared towards productivity maxing.

[C]

Yeah, but that's like, that's an artifact of the generation, is what I'm saying is that, like, this generation is obsessed with. With token maxing. Productivity maxing, like, looks maxing. Like it's all about, like, what can I optimize, right, guys?

[A]

Amir should jump in, by the way. You're maxing. You're maxing this convo and with. You're doing it without him.

[D]

But I think it's really important because I think this is not. I think it's really important. It's not just a question, Dave, of like, do you build a better UX on the thing? It's like, what are you even building?

[C]

Well, just like, I guess that's what I'm saying is I both agree with you, but I also think that computers suck and they waste an enormous amount of people's time in today's world. Like, I'm not talking about AI. I am talking about the Mac that you are talking to me on. I'm talking about the iPhone that you use or the Android that use. These things are awful. They take an enormous amount of time and energy away from people that could otherwise be going towards human relationships, their friends, the work that they're actually trying to do. Like, oftentimes pen and paper are better than trying to design something on a computer today. And so I guess that's to me, like, the great thing about AI is that we can now just talk to computers with English. And we're still trying to figure out how to do that in an effective way and help everybody else do that in an effective way. But like, to me that's kind of the, the thing we're trying to get to is that computers are no, no longer suck in our lives and waste huge amounts of time. And it's not even about productivity maxing. It's actually about getting the time that is being wasted out of people's lives and making computers a more productive citizen in society. And that's like the exciting thing about AI to me, it's not the productivity maxing thing, but the way we have to get there is through design and user experience and like improving the way that this stuff works. And so anyway, that's like kind of my take on it. And Amir, you should talk.

[B]

I agree with a lot of what was said. I always think back to the question of how big is a market. I'm sort of responding to Jessica's earlier comments. And for a long time people were like, oh, you know, Google's best days are behind it and it's not going to grow, it's not going to accelerate growth. And the market is X big and it was like 10 times bigger than anyone thought. I think it was the same for a lot of companies, if I'm not mistaken. Jess, you wrote a whole column about this seven years ago. And so I think in this case, just like Dave was saying, and Brit too, we are living in the future here in Silicon Valley. We're like at 1% penetration of what is the market in terms of what you can currently do with these tools. Only 1% of the people are actually doing it. We're still only like 12 or 24 months out from like past, you know, the point where you were still telling people, oh, you can put your blood work into this and it will actually give you extremely useful information. And I realize, Jess, you're saying not everyone needs that or wants that, but people still have no clue what they can do. So bridging back to Dave's point, there is still a discovery problem. It will be solved over time, I'm sure. Maybe it's already being solved, but the market is enormous.

[C]

It's going to take years, man. It's going to take like literally 10 years.

[D]

I'm not anti. I started by saying AI is going to transform society and how we live and work. Right. I think the market for these like all encompassing personal agents is very unclear. And I think the, the sort of mentality of like, look at all the things that I do as a power user, therefore the market size is X is not actually how this technology rolls out. I mean, what percentage of people in the America, the United States, even use Google Docs?

[C]

Sure. A lot, actually.

[D]

But Jessica, I'm not, I don't think so. Actually.

[B]

You're talking about agents. I view these things right now. I'm not in like agent land like the rest of you might be, but I still view these things as at a friend who can like, help solve problems, can give you ideas for how to solve problems, whether it's being your car mechanic or plumber or anything else, or doctor and to just help you measure things that you never could before. And we're, we're in this environment where a lot of white collar industries are, you know, shrinking, having layoffs. You talk to people who are left at these companies, they are being forced to think about how to be, you know, more efficient and use these tools like never before. So every conversation, I'm sure you have the same conversations every weekend are very, very illuminating. You have like an advertising executive who's like, I had no idea I could measure all this data that we have to create, you know, a dashboard or create like a report telling, you know,

[C]

go a day without hearing this. Amir, like, it's like amazing. And I think, like, but I also

[D]

hear, yeah, I built a bunch of stuff and I'm good for now.

[C]

I guess your job is to be skeptical.

[D]

Amir is way more skeptical than I.

[C]

But, like, we're also running, we're on an super insider podcast. And like, part of what we do in Silicon Valley is work through these Things, right? Like we talk about them to try to figure out like how to take this technology further. How do we make this tool better, can we fit it? And sometimes the technologies don't fit, right. Like crypto tried to fit itself into a lot of different use cases that it turned out to be absolutely terrible for. And like normal databases were just fine. And that might be the case for agents too. But I'm also hearing every single day the same thing that Amir is hearing, which is like literally every day I have somebody, usually they're like pretty normal, small business owner that's like, oh my God. I didn't know that like I could load in 20 years of legal files into an agent. And like I've been trying to like literally this weekend I was on a dad's trip in Zion with UTVs and there were some guys I never met before and this guy's like, doesn't work in tech and he's just like, I've had this 20 year legal problem that I could never solve. And I uploaded all of my legal docs and all the emails for from 20 years of negotiation. And I'm literally going to be able to solve this with the counterparty next week on Tuesday because we can now finally see where it all connects. And it's like, it's like reducing the confusion. And like that's amazing for him gotten

[D]

a point of this con, sure, that's great. And he'll come up with a bunch more things and he'll use increasingly open source models. But he is a lawyer. That is a very specific.

[C]

No, no, he's not a lawyer. He's just a. He's just a real estate. He was. He's like a real estate professional that like works in commercial real estate. And he had this like this deal that had gone on for, you know, it was a local deal down in Southern California, like in the, you know, counties around San Diego. And like it's just, you know, normal, normal work. Right?

[D]

Yeah. I just don't think that moms across America are going to have personal agents. I don't, I don't. I think.

[A]

Okay, what are we waiting?

[C]

Just like what are we waiting? Yeah, you need to make a bet on this.

[A]

What percentage a year from now?

[D]

I can't make bets, unfortunately.

[A]

Amir, you want in?

[C]

Why can't you make bets?

[A]

Because you guys are journalists. What? You never use Polymarket.

[D]

You have to go soon. We're moving on time. But also like it's a pointless bet because you'll just redefine agent, right?

[C]

No, Agents are gonna. Agents are gonna be the thing for a long time, Jess. Like, it's gonna take. It's like we're gonna be talking about agents. Like we talk about computers. You know, it's gonna be like, computers

[D]

is Gmail adding something to my calendar and agents agent, or do I have to be using Spark or what? I mean, it's just too. My just fundamental point is like, we cannot assume that the goal is more productivity for everyone. I agree there are fields and people where it is, but this technology is being designed by people who think that. And I think that could be its undoing. What's going on, Br? You got any pop culture? I got a book recommendation.

[C]

I just drove 400 miles through Zion in a UTV with Tell Us More. You know, I've never done this before and I had never even considered doing it, but a friend of mine does this dad's trip every couple of years and invited me to go. And it's like the most dusty, dirty, crazy experience. But these four person UTVs are kind of a magic technology. Like you can go 50 miles an hour across the desert in super bumpy terrain and you barely feel the terrain. It's like you're floating. And it was a really cool experience that, you know, is like pretty accessible. You go to Zion, go to the national park, rep one of these things and go camping. You know, we pitched our own tents. It was a great experience for my boys. I kind of couldn't recommend it more. It's very challenging, very difficult. Like, hard to be, you know, full of dust every day. But, you know, it was really neat.

[A]

So Dave went offline very often?

[D]

Yes.

[C]

Oh yeah, they actually, they also took our. They took our phones away and made us put our phones away for the whole weekend. So everybody was completely off the grid. And it just reminded me, it took me like the whole first day to stop thinking about some things I had been reading on the Internet. It's just, you know, we all talk about this all the time, but there is really something about truly getting off the Internet and detoxing from it. And I think we all talk about it, but don't actually do it. And so I was thinking about that a lot that, you know, we've talked about it a lot on the pod, but now that you can generate anything you want on the Internet for free and that, like, the Internet's just like full of infinity content, I do think actually these offline, truly offline experiences, like, are going to become extremely valuable. Like, I don't know, I could not recommend it more. So it was great.

[A]

I'll bridge from that back to the online sort of, which is, you know, there's been a lot of news this week with AI entertainment, so. So Suno raised around like 5 billion. 11 Labs has these. All these new launches of new music verticals where the track can actually change genres mid song. So you can, like, start listening to a song with pop and then it moves to rock, then it can go to, like, country in the same, with the same lyric base. And eleven Labs is also resurrecting Stan Lee's voice and a partnership with Marvel, which is, you know, kind of the second big one after Val Kilmer has licensed his voice for movies. And. And then Steven Spielberg, you know, Cannes happening over in Europe, and Steven Spielberg is like, warning Hollywood, like, guys, you have to make original stories or like, this industry's toast. And. And so I think that, you know, Steven Spielberg saying it is like Steve Jobs saying it and tech. And. And I think that, you know, we. We all know this is happening to the entertainment industry, but it's like the people are finally, I think, trying to take it seriously. And yet there's this binary thing happening where, like, AI Canva is happening in Suno and 11 Labs world. And so I'm. I'm fascinated by the sort of the binary nature of AI Entertainment and original films. And then I'll end with two original films. I did watch this weekend while Dave was offline and I had two nights to myself. One was called Remarkably Bright Creatures, Sally Field on Netflix. So good. And the Other is an A24 film on Apple TV called Eternity, where this woman had two different husbands in her lifetime and then they all die and she has to choose who she's spending eternity with. I think you'd like it, Jess.

[C]

You better choose me.

[D]

I have one recommendation in a story, but Amir, do you my story?

[B]

I want to hear your Mark Rober story. Yeah, so tell the story.

[A]

Are you meeting him? Do we know him? Is he coming on the pod first?

[D]

I highly recommend a story called Yesteryear. Now, this is a novel. And when I heard what this novel is was about, I did not think I would enjoy this novel because it's about a trad wife, of which I am quite far from one.

[C]

I don't know, you might be kind of a Silicon Valley trad wife.

[D]

You know, guys, I got no sour a friend.

[C]

No. What that means in Silicon Valley is that you're working. You're like, you know, that's like the way you do it around this town.

[D]

I don't even think there is a way anymore. But that's actually this book. So I was like so skeptical, but I saw it and I was reading the information summer books recommendation, I was triangulating all these lists. I'm like, yesteryear is on all of them. So I start reading it and I cannot put this book down. I'm planning my evening in my head and I'm like, I'm going to read some yesteryear. Like I am right now. Because, you know, I'm so engaged in this podcast. But it. So the premise is of course a trad wife who is a huge influencer with a huge following. But the twist, which sounds corny, but I promise you it's not, is that she gets transported back to the 1800s, but it is not weird. And I really have felt that this book captures like the experience of being a woman better than anything else I've ever read. So I highly, highly recommend it. Secondly, yesteryear. And I can't reveal too much because I. I will be writing about this. But I did get to go to Crunch Labs yesterday and meet the amazing Mark Rober and spend some time with him. And he is wonderful. For those of you who don't have parents in school, Mark is an amazing YouTuber who has kindled kids love of science through his very, very wacky, outrageous experiments that are crazy. He's been doing one video a month since I believe 2015. I first upon hearing that at the end of the day, well, like they were waiting for parent pickup. They showed our kids Mark Rober on YouTube like had a mini conniption that they were like throwing up YouTube videos waiting for the bus. And then after watching them, I kind of like got it. So anyway, I had an amazing experience which I will be writing about. But for those of you who know, one of the things if you visit Crunch Labs, which is top secret so I can't tell you where it is, is you shoot a tennis ball cannon and if you hit the bullseye, you get $10,000. Four people in history have done this. No, you didn't until yesterday when I hit the bullseye. Editors.

[A]

All I do is win the video.

[D]

Please. The video is forthcoming because we have the video of my reaction to which Mark just like falls to the floor and it's hilarious. We will get the video of the shot. For those of you ready to like wring my neck about journalistic ethics, I will be not, I will not be accepting the money, so have no fear. But it was really, I'M sure. I'm sure I can donate something awesome. I'll work with the crunch lab teams to find something I don't know that would create more problems.

[B]

We're trying to write about. We write about open cloud. We can't. We can't contribute.

[D]

But anyway, I. It's kind of cool. I mean, it's obviously, like, completely dumb luck. Although I have taken a few, like, shooting classes, and I did remember, like, you, like, pause your heartbeat. You take, not pause your heartbeat. You, like, tune into your heartbeat. Take a deep breath. So I had, like, some fundamental skill. My kids think I am the coolest person in the world, which is important. And number five in history. You're looking at her. We do not need to plug Mark Rober on this podcast, but more cool things coming for him. And that was, like, a pretty wild experience. And just the nicest guy who really has the stoke for kids loving problem solving and science. It was great. So, anyway, okay, guys, nothing happened on this pod. No fireworks, no energy. Same old topics. I think it was a mirror. Amir, we're gonna have you back.

[B]

I hope you didn't feel like we ganged up on you. Jasmine, I hope you.

[D]

No, I. I can handle it. Believe me. I think my husband's gonna be very proud of me, so that's what I'm in it for. And it was great to be here with you all. So we will see you again I. Next week. I might have a conflict. I might not. I have to look at my calendar.

[A]

But maybe Amir will come back some company if.

[D]

If I have the conflict. Amir is the conflict. But some combination of friendly faces will be here next week for maybe Mark Rober, maybe. No, not that. No, it's happening for another episode, more or less. Bye. Bye, guys.

[B]

Bye.

[D]

Bye.

[A]

If you enjoyed this show, please leave us a virtual high five by rating it and reviewing it on Apple Podcast, Spotify, YouTube, or wherever you get your podcast. Find more information about each episode in the show notes and follow us on social media by searching for or less avemorin, lessonlesson. And as for me, I'm Brit. See you guys next time.