Introduction

I would like to talk to you about Emad Mostak's The Last Economy. And basically his kind of premise of maybe in a thousand days, give or take a couple of thousand, AI might rewrite the rules of work and value.

And we've heard from David on some insights around law.

Who Is Emad Mostak and What Is the Intelligence Inversion?

Who is Emad Mostak? So he's a stability AI founder, so similar to ChatGPT providers OpenAI or Anthropic.

He's a former hedge fund manager, and his central premise is that within 1,000 days, AI triggers an intelligence inversion. What does that mean?

At a very high level, machines become better than humans in all cognitive domains. Knowledge that is currently scarce resides within a lawyer's brain, suddenly becomes non-scarce and available to everyone. And at that point, capitalism's whole premise around scarcity collapses. Human labor, well, that suddenly has negative economic value compared to currently AI at one penny per query.

So, Emad's been described by some people as a visionary and some people as a very smart charlatan with a galaxy-sized ego. That from a friend who's worked with him. I also know that one person in this room has described him as an eloquent Froot Loop, which I also like.

How The Last Economy Is Being Distributed

So before we continue, I just want to take a moment to look at how Mostak's actually distributing this book.

So we've got like the usual kind of download a PDF, an EPUB, you know, get it on Kindle. And then we've got these AI kind of things around Notebook LM, that's from Google, OpenAI's ChatGPT5, Anthropix, Claude, X.AI, that's Elon Musk's, Grok.

And then this thing, GitHub, which for the technical people means he has a repository. He's uploaded the whole book to the repository. For the non-technical people, it just means that you can do a bunch of stuff with it in technical ways.

The best thing here is these reviews created, five-star reviews created by the AIs. They obviously love Emma Mostak.

So anyway, I'd be really interested to hear of anyone's views on that distribution over pizza.

Defining the Intelligence Inversion

The intelligence inversion. So what is that?

The current state is that humans create value. Apparently, what we're doing here right now creates some sort of value to someone. Machines augment us. So I use AI in the production of this thing.

Labor scarcity drives wages. So the scarcer the kind of skill, the higher the wages.

1Emma's prediction is that this is inverted once AI exceeds human capabilities by having all this knowledge. So machines exceed humans. Humans augment AI. And labor is negative value.

So we're going into kind of matrixy kind of area here.

And what's important is that he doesn't see this as a gradual transition, like the Industrial Revolution and steam engines coming online and gradually being built out and rail networks. But he sees this as a phase transition. So suddenly changing from water into steam. So a relatively sudden shock.

And in his view, that is a thousand days, but it's in order of magnitude. So not 10,000 or 100,000 or a million days, but he thinks somewhere between a thousand to 7,000. So, you know, 8,000, whatever that might be. That's a few years.

Seven Fatal Lies of Current Economics (According to Mostak)

He discusses the seven fatal lies of current economics. And I've put there, as an asterisk, according to Mostak. So this is where we can potentially critique it.

So he says, well, scarcity is fundamental and that is like critical to traditional capitalism. Bearing in mind capitalism has been around for about 250 odd years and it's the dominant economic system that we have now.

It's not been around forever. So maybe keep that in mind when thinking about these fatal lies.

Scarcity is fundamental, but what happens when AI creates an abundance, especially of knowledge? Is the question that he asks.

Human labor has value.

Well, what happens when an AI can do a lot of knowledge things that we do better than us, faster than us, you know, not require food, sleep, a nice cuddle.

GDP measures progress. So we always track GDP and it's held out as this ultimate metric that we must aspire to. However, he believes that it optimizes transactions, not flourishing.

This is a big one for him.

Markets, Monopolies, and Growth

Market self-correct. That's a big one that capitalists hold but it potentially ignores monopoly dynamics so if we think about everything ai currently being supported by nvidia chips largely there's amd and others but by and large nvidia is the largest company in the world by market value and supports an entire ecosystem

tending towards monopoly dynamics. So what happens when those take hold?

Growth is always good.

There's one called Piketty who says, no, when there's when capital growth, that is just sheer investment exceeds the economic growth, then that leads to inequality. So you might see that in the large amounts of billions, hundreds of billions going into AI right now.

But then on the other hand, we're seeing a lack of economic growth, the have your say forums on BBC, a chocker with 3.8% inflation today and thousands of people saying it's more than that, I'm not getting what I used to get.

So is growth always good? Even if we've got growth, does it necessarily translate into people's flourishing?

Jobs, Labor, and Control of Machines

Technology creates new jobs. Well, not when machines exceed all cognitive work.

I think David touched on, you know, Those 30,000 people coming out of law school, where are those jobs?

And are they as legal engineers? Do we need 30,000 legal engineers and legal like ops people?

We control the machines in the intelligence inversion machines.

become economic actors. And we're already seeing this with MasterCard, PayPal.

They've launched their own agents that can interact with each other and say, hey, should we make this, you know, come to this arrangement? And then they go, yeah, sure. All right.

Can you pay me? And there are examples of agents doing that now.

So what happens when they become intelligent enough to just be autonomous?

The Metabolic Rift: Labor Without Bodies

Mostak talks about this thing called the metabolic rift. So humans, and this is all about the premise of labor. So labor comes from use of our muscles, our brains that are sustained with food and drink and all those other things, pizza.

What he's saying is, well, suddenly, AI doesn't have those same needs to produce at least the knowledge labor. And we're not even at the point of talking about robots, which are actually making pretty significant advances as well.

So AI apparently has no physical needs, but I'd say, what about the water, the data centers, the energy, like all of those things? There are material elements to that.

But that aside, the costs are trending towards zero for more and more intelligent tasks. They, generally speaking, do improve with use. And they take hours to train versus years to train.

I mean, we've trained for how many? Coming on for decades to be lawyers.

How long does it take to train Harvey in a new area of law?

Early Signals in the Market

This rift is kind of already happening. And these figures, they're more about the use of AI and how it's starting to result in an uptick in use.

So going from $1 million to $100 million, this means annual recurring revenue in eight months with just two people, two founders. Now, I think it took...

Instagram, like back in the day, a year to get to like a million per person and they had eight people. So that was eight million in a year. So AI is accelerating this ability for people, small numbers of people, to produce things that are used by a large number of people that results in this revenue growth.

Anthropic that creates Claude, they went from 1 billion in revenues to 9 billion in annualized revenue as of this month. Those are staggering numbers. I can't even comprehend them. But basically, it's a huge use in AI.

Three Potential Futures in ~1,000 Days

So what does this all mean in a thousand days? What is he actually saying? He's saying there's three potential futures and we may already be on the path to this already happening.

If we keep on going as we are, maybe there's a great fragmentation. Nations retreat, there's an AI arms race, protectionism, resource conflict fragmentation. I feel like that's kind of where we are geopolitically already.

It's scary to think that AI could potentially enhance us towards that future.

Could we move towards digital feudalism and a comfortable cage? Maybe that's the matrix. I'm in the matrix.

I like to have my stake. Don't let me out. I don't want the gruel.

The AI owners then become the AI aristocracy. That doesn't really roll off the tongue. But you get my point.

The people who own it, so like the Sam Altmans and the Elon Musks who own the technology and the people who are investing in all of the capital growth, those are the people to whom the gains will probably accrue.

And the rest of us, do we then depend on universal basic income or something else? Our agency is eroded, our dignity is apparently lost.

Toward Human–AI Symbiosis

What Mostek is arguing for us to aim for is human symbiosis, living in partnership where human and AI collaborate and we get value from creativity, judgment, care, not just productivity. touched on something interesting earlier in his talk about it's not just doing the emails and everything else. It's about that human connection and managing the client and those kind of things. Is that where it moves towards if the knowledge layer is all completely taken care of?

Now, David,

Framing It as a Thought Experiment

I would say this isn't a new concept. A lot of these things aren't new. They're perhaps just presented in a slightly different packaging.

And he calls it a blueprint, but I like to think of it more as a thought experiment, because there's so much to attack in what he says, because there's a lot of concepts that are perhaps not fully thought out. But we'll come to that. But I find it fascinating as a thought experiment and ideation.

The MIND Framework: Beyond GDP

He's got this mind framework that is an alternative to the GDP failures of which measure transactions, not flourishing, optimizes for the wrong things. For example, divorce. Not great for human flourishing.

But from a gross domestic product perspective, hey, we've got two households, two houses, two rents, two mortgages, two sets of legal fees, GDP's up. That must be a good thing, right?

So he proposes something else, his mind framework.

And it's, yes, there is material capital, which is important, intelligence capital, which is the ability to take stuff and have it almost as a context and memory so that we solve these problems and can kind of pull from it, a wisdom of sorts.

Network capital, which you might know as social capital, diversity capital, which you might think of as kind of building resilience, coming up with mitigations and having diversity of thought, of different approaches.

Key question, how do you measure this? And he doesn't really go into it.

So one of the really important things here is that he sees this as multiplicative. So if any one of these things is zero, the output is zero. And he says, unlike GDP, which is additive, it just adds on top. So that's one of his big concepts.

Yeah, I said that.

From Theory to Action: Nucleation, Not Boiling the Ocean

Moving from theory to action is maybe something that he doesn't cover as well as he should.

However, I really liked the analogy of it being about nucleation. So starting, you know, if we imagine super cold water below freezing, or below zero degrees, if it's completely pure, it doesn't necessarily freeze. It needs some level of impurity, a tiny imperfection to catalyze a reaction that then leads to a complete frozen ice sheet or a block.

1And from that, he's saying, don't redesign everything. Don't try and boil the ocean, start small.

Historical examples, he gives things like the Renaissance starting in Florence, not the entirety of Europe, or the scientific revolution starting with the Royal Society. So demonstrate viability and then spread.

And I'll actually come on to this in terms of your personal aspect.

Negative Nucleation: Fewer Hires, More Revenue

a negative new case we've talked about jobs and stuff i'm gonna skip over that but the main point to take away is revenue count is up head count is either staying flat or going down so revenue for per employee especially with those exposed to ai is three times higher revenue growth per employee so there is less incentive to hire more people

Positive Nucleation: Local Builders and Talks

Positive nucleation.

There's talks next month. Glenn at Fueled. He's been profiled by Google.

There were only two engineers. Emmanuel, there he is, is one of those engineers.

You know, an app and business built on AI entirely like... It's crazy. That talk's going to be absolutely sick.

Johnny and AI agents, so that's on personal assistants. I kind of did want to go into and show it a little bit here, what I've got, but the tech's not working. And then

Mindstone Bristol, the AI curious of you, sharing questions and skills. I think this is just going to become a build into a virtuous cycle. So key point is local builders with the right tools and focus effort.

Does This Lead to Symbiosis?

Does that take us towards human AI symbiosis? Question.

Critical Analysis and Open Questions

There was a critical analysis which I've kind of talked a little bit about.

It's got American biases. It underplays material constraints, perhaps lacks scientific rigor, underdeveloped solutions. I put that productive critique, not a demolition.

But the core question remains, despite these gaps, and they are real.

They are massive. Huge. They're gaping.

If Knowledge Work Becomes Abundant, What Should We Do?

Even the skeptics agreed in our little book discussion, if eventually AI makes knowledge work abundant within 1,000 days, within 7,000 days, hell, even 100,000 days, what should we do?

Especially starting now. Not in panic, not in denial. but with clear-eyed preparation.

A 30-Day Starter Plan

So I want to give you in the next 30 days, similar to the nucleation, don't read as I know everything, start small.

Three actions, map and automate your tasks, according to these three criteria is it low risk if it goes wrong is it highly repeatable is it amenable to your oversight can you stop it if it's going wrong and then if yes automate them you can do this with a piece of a pen and paper just write down all of your regular tasks you know and then go okay well do i want to be keep on doing that do i do i value it if not

Get it done. And then you could do these things with your teams.

Like in a 90-minute workshop, you could probably list out a bunch of things that you really don't want to be doing. Yeah, I've had some interviews with people and they're like, just like mind-numbly boring. I want to shoot myself. Those kind of tasks. If you don't want to do them anymore and they meet the three criteria.

Build Your Personal AI Stack

Build a personal AI stack.

Someone in this room has said previously, well, I don't know where to start. Well, just pick one. I'd suggest Cloud Pro.

Get really familiar with it. And then once you've done that, Add one or two new tools based on your needs, not everything.

Maybe it's whisper flow. I do 95% of my input into my laptop and my phone is voice. I don't type anymore.

I hate it. I hate being forced to type. It's so slow.

My partner who's also in the audience sometimes goes, I don't know whether you're talking to me or the computer. That's a little bit strange, if I think about it.

Advance Only After Habits Form

Advanced integration.

Once the habits form and the needs are identified, then move into more technical things. Johnny's going to be doing Cloud Code and Obsidian and all kinds of things, which are basically building personal assistant agents. And Johnny will show that in November.

And there's these acronyms. We love acronyms. So model context protocols and APIs. They're just tools.

You don't need to worry about them right now if you're not at that stage. And then you can get into coding agents that you don't actually need to use for coding. You can use them for non-technical tasks.

Invest Now; Compounding Will Follow

Your return on investment will be much higher than Netflix. Don't worry about 20 to 40 pounds a month. You'll...

Start off small and then eventually you'll be saving or generating a return on investment that is five to 10 to 20 to 100 times higher. Invest now and you'll get the returns. It's compound interest.

Learn Locally, Iterate Together

Okay, local learning, this is what we're doing here.

And get together, try these things. Learn what works before you invest and adapt to proven workflows.

Conclusion: We Are Future Cats

Okay, this is a fun slide, but basically this is the closing slide.

We are future cats.

Imagine a cat looking at a red laser dot, pen dot thing, and imagine that it knows, okay, that dot, I chase it. And then imagine you can probably go, okay, so when they push the button, that means there's gonna be a dot. know what the laser pen is and they might be able to connect that when you push a button it results in the dot but they have no clue about the battery the electronics whatever else we are this cat we might be able to expand our horizon slightly but there are certain things that we just don't know so

Keep an Open Mind, Prepare Regardless

maybe scarcity and capitalism that is based on it and which we kind of are surrounded by people saying this is the way it's always going to be maybe that's just like here and what's coming is out here so with that in mind just keep an open mind and prepare as if

it might be happening because if my stack is 30% right, you've prepared. If wrong, you've still improved.

Discussion and Q&A

Let's discuss over pizza. I think I'm going to leave it there and we can do questions over pizza.

Hands up if anyone really, really, really would like to ask a question now. Oh, you're the one who's in the way of pizza now.

Should We All Become Plumbers?

Does this mean we should become plumbers? Yes. And electricians.

And enjoy all of the physical things that we do. A lot of us might go to the gym or do art. There are people who are stronger than us that can do art better. But it still gives us value.

Did we? Okay. Yeah. Brilliant. Okay.

Closing Notes and Thanks

I want to basically, there is another slide in here, but it's the feedback one. And I think I can't bring it up right now.

So with that, I want to say thank you to everyone. I want to run the lottery, which I'm going to do now. And you won't be able to see it. But basically, it will give the whoever wins thing.

Oh, no attend error. No attendees have attended the event.

This is what happens when you build a platform on Replit and put it out into production. I'll figure that out and we'll sort it out later.

Thank you for your time and let's go and enjoy a lovely conversation over pizza.