The Cradle of Civilization Gets an Upgrade
This article is a possible vision of the future. A hopeful one. It combines and extends AI * blockchain into a symbiosis of biology and machine, for a future of abundance.
There’s an old joke from the 90s that “on the Internet, no one knows you’re a dog.” (Of course now we’d say doge!) That joke has a newer incarnation: that “on the Internet of Things, nobody knows you’re a toaster”.
The jokes point to a subtle fact: network technology provides a disintermediating layer where you may not know what’s on the other side. And, you may not care. Or, the other side might surprise you. For example, it might be a half-plant half-robot (a Plantoid)!
These artifacts may sound like science fiction but in fact are emerging thanks to improvements in blockchain and AI technology.
Why stop at self-owning forests? What about a wind farm that owns itself? And solar farms? And power generation facilities in general? What would that mean with respect to energy?
Here’s a broader question that we could all ask: how can we transition from a mentality of scarcity — a zero-sum game of “I win, you lose” — to one of abundance — a positive sum game, where “everybody wins” ?
It turns out that there’s a collection of ideas and technologies, that if brought together in a new way, collectively promise not just more efficiency, but potentially an upgrade to the cradle of civilization itself. I call it Nature 2.0.
The next few sections will be a series of building blocks towards this vision. I will cover Nature 1.0, blockchain, AI, and AI*blockchain (AI DAOs). Extending beyond, we get Nature 2.0. Abundance included. Let’s get started!
Take a tree. A tree takes in nutrients from the nearby soil, rain and other moisture from the nearby air, CO2, and sunlight. It uses that to grow — to manufacture wood and leaves — and to output oxygen and sugar. Importantly, all these operations are local.
This tree is in symbiosis with its local environment — with the squirrels, the earthworms, the grass, and so on.
Given all this interaction, it’s actually remarkable this tree don’t have an identity. It doesn’t know who it is. And it doesn’t care. It doesn’t have a concept of ownership; it doesn’t have an idea of scarcity; it doesn’t have an idea of money, or value. It just is.
It’s self-sustaining (within its minimal biosphere): if you chop off a branch, another will grow back. It spreads — it grows seeds which fly through the wind, and launch new child trees that grow wherever the seeds land.
Anti-fragility is not just about resilience, but actually getting stronger when you’re down. In some ways, trees are anti-fragile too. For example, a forest fire (a kick) makes the forest of trees stronger for the long haul.
Nature is not just trees, of course. With nature, we go to ever-higher levels of integration, starting with the flora and fauna — from plants and beasts to larger and larger ecosystems. Forests and valleys, islands and continents. Each ecosystem just is. Each ecosystem is self-sustaining; and many are anti-fragile.
And what if you go to the very largest ecosystem? That is our planet: Mother Earth. We as humans and the ecosystem around us live on substrate that sustains us and nurtures us. It is our cradle, our cradle of civilization. Without it, we have nothing. This is Nature.
Yet, we’re falling short. We have scarcity. We’re competing with each other in places we don’t necessarily need to be, from poverty to famine to global warming to disappearing species. We don’t have abundance.
Yet — to ask courageously — perhaps we could get to abundance? To follow up, we can ask: in today’s day and age, does tech have anything to say about catalyzing abundance, and reconciling it with Nature? This essay poses these questions.
To get started, I will review two powerful technology levers that promise to make a difference: blockchain and AI (artificial intelligence). Then, I will synthesize them into AI*blockchain, then extend the ramifications towards these questions of abundance.
Ways to Frame Blockchain
Let’s start with blockchain technology. I’ll supply many framings, starting from “mundane” low-level definitions up to high-level societal ones.
“Chain of Blocks”
The most basic framing of a “blockchain” is: it’s simply a data structure in the computer science sense. The data structure is a chain of blocks where each block is a list of transactions.
A transaction can be to issue an asset, or to transfer it. To issue means saying “I’m generating some Bitcoin tokens, or energy tokens” where “I” might be an external user or the network itself. A transfer means moving the ownership of asset from person A to B. These transactions are digitally signed making them super secure.
Once you have those lists of transactions — those “blocks” — you simply link them together in a secure way. It forms a “linked list of blocks” where a link is a pointer to the previous block’s hash. Then you let anyone keep a copy of this list, where each person’s copy automatically syncs with others’ copies throughout the planet. That is, the lists are in “consensus”.
“Database With Blue Ocean Benefits”
A framing for “blockchain” one level up is: a database with blue ocean benefits of decentralized, immutable, and assets.
Let’s start with databases, as in Oracle databases. A database is simply a way to store structured data and query it efficiently. Below is a picture of Oracle HQ, which mimics the icon for databases (lol).
Traditional databases are administered by a single sysadmin. They are “distributed” if they store the data across many boxes. But, one sysadmin could delete the whole system — oops!
A blockchain is a database that adds new “blue ocean” new characteristics:
- It’s decentralized — no single entity owns or controls it. There’s no single point of failure.
- It’s immutable — once you’ve written into it, it’s there for good. (Practically speaking.)
- And it has assets,— you own something if you have the private key. This can be for something digital, or even physical through “digital twin” identities. This article elaborates.
“Smart Contract Platform”
We framed blockchains as structured data storage platforms (databases) that are decentralized. We can also frame them as processing platforms that are decentralized. That is, “smart contracts” substrates. Conceptually, we add a virtual machine (VM) to run the smart contracts, and use a decentralized database to store that VM’s state.
I’ve described how we can frame blockchain as decentralized database or processing. We can generalize. In the past, we’ve had compute stacks for the desktop, the early web (LAMP stack), the cloud (eg AWS services), mobile, and so on. Now we have an emerging compute stack for blockchain — a decentralized compute stack.
We can organize the stack according to the three elements of computing: storage, processing, and communications. Within each, there are a few building blocks. In storage this includes file systems, databases, and stores of value. In processing this includes smart contracts, high-performance compute. And so on. This article elaborates. So, “blockchain” isn’t just one magical monolith, it’s actually a whole bunch of building blocks.
Let’s go to a higher level yet. People call blockchains “trust machines,” because they minimize the amount of trust you need to operate.
I will illustrate with an example of remittance. If I want to transfer money from a bank account in Canada to an account in Germany, it would have to go through 2–3 intermediate banks. Each step costs $10; you’re paying for trust in that node (based on size, age, etc). And it takes days to clear.
Blockchains give higher confidence in the transactions, for pennies, clearing in seconds or minutes. They do this via cryptography & consensus, rather than reputation. Put another way, as “trust machines” they minimize the trust needed (in humans/banks) to operate. In doing so, they allow larger and larger organisations of people to interact without needing humans in the middle. Nick Szabo calls this “social scalability”.
Sometimes “trust machine” gets interpreted as “you can write data to a blockchain and suddenly the data is truth.” This is incorrect. Blockchains are not truthwashing machines.
“Magic Internet Money”
The value of a given currency is based on the shared belief of its value by the holders of that currency and the ecosystem around it. A trust machine makes it way easier to keep track of “who owns what tokens”, and by extension, the shared belief of those tokens’ value.
If you asked me what the most important characteristic of blockchains was, I’d say “incentives”. Why: You can design a network that gets people to do stuff, by rewarding them with tokens.
Bitcoin is a great example. The Bitcoin network aims to make it hard to unroll its list of transactions — its definition of security. It rewards people, in Bitcoins, if they add to its security by adding hash rate which is basically electricity (assuming ASICs). Effectively, Bitcoin is maximizing electricity.
How well does the incentive machine part of Bitcoin work? Amazingly. It’s using more power than most small countries, and is on track to use more electricity than all of USA by mid 2019. Now, you might think: that’s terrible! I agree, that part is terrible. However, there are now much more efficient blockchains than the 10-year-old tech of Bitcoin, so please don’t dismiss blockchains based on Bitcoin’s inefficiency. My point here is to illustrate the amazing power of blockchains as incentive machines.
“Public Utility Network”
We have waterworks and electric grids as public utility networks (PUNs). Now we have ones that are more information-centric. Bitcoin is a PUN too. It’s a service that’s out there for all of humanity to use. In its case, it’s as sort of a store of value or a digital cash. Another example is Ocean Protocol, which is a PUN for data itself — for buying and selling data, towards a global data commons for all of humanity to use.
These public utility networks are self-sustaining: they’re really hard to kill. To kill Bitcoin, you’d actually have to take out every single one of those 6,000 nodes. And there’s always going to be a handful of 10 people (or maybe 10K!) that will run Bitcoin nodes into their grave. These networks are anti-fragile too. If every time there’s an attack on Bitcoin, if somebody actually managed to attack it, other people will come in and fix those bugs, so every single attack makes it stronger and stronger. With self-sustaining anti-fragile characteristics, PUNs may even achieve 10,000 year horizons, Long Now style.
PUNs aren’t limited to giving IT services, aka things with bits. They will also relate towards things with atoms. This post elaborates.
“Decentralized Autonomous Organization”
Let’s go to a higher-level framings yet: blockchains as “Decentralized Autonomous Organisation (DAO)”. A DAO is a computational process that’s running autonomously on decentralized infrastructure. No one can pull the plug, not even Bezos (who could pull the plug on AWS). And it’s got resource manipulation, including a wallet to hold Bitcoins, Ether, whatever.
A DAO is autonomous code that can own stuff. It’s a “computer virus” but benevolent. Bitcoin is a DAO; so is Ethereum. The more autonomous applications running on top of Ethereum are DAOs too.
Erwin Schrödinger framed life merely as physical processes in his treatise “What is Life?”. More recently, physicist Jeremy England has given it a thermodynamic framing: it’s all about entropy. Carbon is not a deity.
In this vein, Ralph Merkle has called blockchains new life forms on the internet:
Ways to Frame AI
I’ve given several framings for blockchain; now I will do the same for AI. Then, you’ll see when we combine blockchain with AI, sparks fly.
“Replicates Human Cognitive Behaviour”
One of the original goals of AI is to mimic what a human does. It’s best exemplified by the Turing test: deciding whether someone behind a wall is a human or a computer as they pass messages back and forth. While interesting, the AI world has found that there are more useful definitions for practical settings.
“Can Do Tasks That Only a Human Could Previously Do”
Here’s an example. The image below shows an antenna evolved by NASA researchers Jason Lohn and Greg Hornby. For its application, outperforms all human-designed antennas — and it’s basically a bent paper clip.
“Can Do a Task at a Speed / Accuracy / Capacity Not Possible by a Human”
Here’s an example. A modern chip may have 10 billion transistors, designed by a team of 20 humans in just a few months. It’s possible thanks to AI under the hood of CAD tools used by human designers [McConaghy et al 2012].
“Set of Tools”
You can also give AI a mundane framing: it’s AI is simply a set of tools like classification, regression, knowledge extraction, optimization, creative design. The image below has an example, which inputs a 5-dimensional raw dataset and outputs a decision tree for making circuit design decisions [McConaghy et al 2008].
Going more general, there’s a sub-field of AI called Artificial General Intelligence (AGI). And it’s really about embodied agents, whether they’re robots running around in the physical world, or in a simulated virtual world. And this is a very young field, it’s not that far yet, but it has some very strong ideas about where the world can go.
AI * Blockchain
The last two sections gave framings for blockchain and for AI. We can combine these into something called AI DAOs: Artificially Intelligent, Decentralized Autonomous Organisations.
What’s an AI DAO? It’s AI running on a decentralized processing substrate; or it’s simply a DAO running with AI algorithms.
On example is an AGI algorithm running as a smart contract substrate like Ethereum, as the image below shows.
The ArtDAO Example
Here’s a simple example: The ArtDAO.
Once the DAO generates the art, it simply sells it, such as on the decentralized marketplace OpenBazaar. It earns some Ether or Bitcoin.
Then, it repeats. Over time, its wealth accumulates. Let’s say it costs $1 of compute power to generate one image, and the ArtDAO sells it for $10. And then the ArtDAO creates 10 more images and makes $100. And then sells those, getting up to $1000, then $10K, and so on. Soon, we’d have the world’s first AI millionaire! No human can control it.
This is possible with today’s technology.
And this is just a start. You could get it to evolve its own code using Genetic Programming (GP). GP might randomly add some code here, and remove some there, and so on. Most changes will be complete garbage. But, every now and then, one will get better. That’s the one that will survive, and reproduce, mutating and crossing over further to have ever-more-interesting behavior. It’s no longer simulated evolution, since there’s nothing simulated about it. It’s simply evolution.
AI DAO Architectures
AI DAOs can come in a few flavors.
- AI at the very center with a bunch of token holders around
- AI at the edges, with some very dumb clearing mechanism in the center.
- Have a swarm of really dumb AI agents, and they are all talking to each other, getting emergent intelligence like Ant Colony Optimization.
Recipes for AI DAOs
There are several ways to construct AI DAOs. You start with a DAO, you add AI, or you start with AI, you put it on a decentralized substrate. Or you start with Software as a Service, which is basically some company running something, you turn that into a DAO, which is basically giving the same service, and then you add AI.
Or you can even take a physical service and turn it into an AI DAO as well. So, these things don’t need to live just on networks, they can actually have bodies attached.
Rights for AI DAOs
AI DAOs can get rights, using today’s laws. Here’s how:
- Start with a corporation, which has personhood and the associated rights.
- Automate that corporation into an AI DAO, by transferring control from humans to an AI running in a DAO. Corporations in Zug, Switzerland don’t need humans attached, as the creators of The DAO showed. The difference to The DAO is that the AI DAOs use AI algorithms for decision making.
And guess what? You’re done. You have an AI with rights. And you can’t pull the plug. We don’t need legislation to make this happen; it’s actually possible today. (For better and for worse. Oh my..)
AI DAOs could be way bigger than AIs on their own, or DAOs on their own. AI gets its missing link: resources; DAO gets its missing link: autonomous decision-making. The potential impact is multiplicative.
I’ve discussed Nature 1.0, AI, blockchain, AI DAOs — how does this all unify?
The quick answer is: AI DAOs to the moon. But this version of “when moon” is positive, I promise! ;)
What does “AI DAOs to the moon” mean?
There is much work on self-driving cars these days. So: what if these cars owned themselves? Every single car is its own agent that with its own wallet. It’s an AI DAO with a car body as a resource attached. (Using DID framing of course.) If the car had to get repaired, it would go to the shop and contract out people to fix it. This is possible. So we would have self-driving, self-owning cars, interacting with the ecosystem around it.
And of course we don’t need to stop at self-driving cars. Self-driving trucks are already in production in the USA. Let’s make them self-owning too.
Furthermore, the roads themselves can be self-owning. Let’s elaborate on this one. Roads get built by taking in taxes, the government building it, and then deploying that capital. It’s about pooling capital and deploying it to usage as a public utility. So we can also pool capital into a RoadDAO that basically buys itself over time. Self-owning roads.
Next up: self-owning wind farms. And, self-owning power grids. This is a nice one to think about because we can inject incentives to be carbon-neutral. This is all possible with today’s technology.
There’s potential for closer relations between biology and machine than first glance. We can bring biology closer to the machines, and vice versa.
- Remember the self-owning forest, terra0? It’s an example of nature bonding with machines.
- We can go the other direction too: bonding machines to nature. That is, take a machine, and make it more nature-like. Plantoid does this. It literally relies on humans to come along and make copies of it, but it incentivizes them, because it gives them tokens to do so.
And, of course, we can have ever higher levels of integration. We can start with the plants and beasts at the lowest level, and go to higher and higher ecosystems; the roads, the power grids, to swarms of these things together. Connected, of course, with IoT, machine-to-machine, all that sort of thing.
Guess what? Each of these cars doesn’t need an identity, doesn’t need ownership. It doesn’t have a concept of scarcity, it doesn’t have a concept of money even. It just is. Self-sustaining, anti-fragile.
And we go to higher levels too. The swarms etc. They don’t have identity, ownership, scarcity or money — they just are. Self-sustaining, anti-fragile.
This is Nature 2.0. It’s biology and machines in symbiosis. Simply providing.
Many dream of Universal Basic Income (UBI), because it can address worries about job loss in the face of AI, and help more people to chase their dreams and self-actualize.
Blockchains make the distribution side straightforward. Anyone that provides some proof of human (even if imperfect) gets an equal amount of income coming to the UBI chain.
The challenge is how to pay for it. Nature 2.0 gives a solution to this! In short: profits from self-owning machines pay into the UBI chain.
Let’s use a self-driving, self-owning car as an example. It generates income via Uber-style ride-sharing. Its costs are just gas and maintenance; it doesn’t have any human mouths to feed directly. So it can generate a healthy profit every month. The idea is that it feeds this profit directly to UBI chain!
Let’s repeat for every self-owning car. And truck, road, and power grid, etc. The abundance from machines gets transferred to humans. (Kudos to JP Doomernick for conceiving of this. Awesome!)
Pros, and Cons
Like many technologies, AI DAOs are a double-edged sword. There is a positive and a negative framing. “Nature 2.0” is a positive framing of AI DAOs. There’s a less-positive framing too, where we inadvertently give all our control to the bots. #ooops. This article and this talk [slides] elaborate. Let’s do our best to ensure the positive version prevails!
Manifesting Nature 2.0
How does this get built? One step at a time. We’re much closer to this than you may think. Here’s the steps from yesterday, to today, to tomorrow:
As the tweet mentions, decentralized substrates to support “agents” are coming. They’re getting built, today. These include my team’s work at Ocean Protocol, as well as our friends at SingularityNET, Fetch.ai, PolkaDot, and more. There’s also great work on M2M (machine-to-machine) economy, UBI, and more. As these building blocks mature, they will start to get connected. Like any good exponential: at first, the transformation will look slow. Then, it won’t.
In Nature 2.0, no one knows you’re a forest.
In Nature 2.0, no one knows you’re a grid.
You just are. While you simply provide.
Taken together, we are on the cusp of an upgrade to this cradle of civilization that we call Nature. Nature 1.0 is seeds and soil. Nature 2.0 adds silicon and steel. It’s biology and machines in symbiosis. Simply providing, towards abundance.
Appendix: Idea Provenance
[Section added July 30, 2018]
After I wrote this article, I learned of writings by some people who had some of these ideas long before me. Cool! Here, I give them their due.
Science fiction writer Karl Schroeder reached out in the comments section, with the following:
where McConaghy ends is where I began 18 years ago with the AI-awakened Nature of my New York Times Notable novel Ventus. The idea of the AI-supported “self-owning forest” was first articulated, as far as I know, in my 2010 novelette, “Deodand,” published in the anthology MetaTropolis: Cascadia.
Awesome foresight, Karl! I have some reading ahead of me:)
Also, Stephan Tual reached out and pointed to his current “Deodands” project is about manifesting some of these ideas now. Amazing. Here’s the video. Stephan points to an influence by Mike Hearn in 2011, and also learning of Deodands from Karl in an Ethereum discussion board in 2014. The circle closes:)
Appendix: Further Reading
- AI DAOs Series, blogs and videos, by Trent McConaghy. The articles give further elaboration on AI DAOs and their implications. Related is this talk [slides] about tokens and complex systems that I gave at Santa Fe Institute in Jan 2018.
- Ventus and other writings by Karl Schroeder. Where this article stops, Ventus could be what happens next. (Thanks again Karl!) [added July 30, 2018]
- Daemon and FreedomTM, a two-novel series by Daniel Suarez. Starting with the present, a bunch of “dumb” agents interacting causes a chain reaction of events that ends with self-driving SUV swarms and a radical restructuring of society.
- Out of Control: The New Biology of Machines, Social Systems, & the Economic World, a non-fiction book by Kevin Kelly. It “chronicles the dawn of a new era in which the machines and systems that drive our economy are so complex and autonomous as to be indistinguishable from living things.” I read this two decades ago and it has influenced my thinking ever since.
- Swarm Intelligence, by Eric Bonabeau et al. A layman-friendly account of swarm-style AIs, and how complex behavior can emerge from really dumb agents. This book launched the field of swarm AI. If you love that, follow up with John Holland’s Emergence. Both have roots at Santa Fe Institute, which is ground zero for Complex Adaptive Systems.
- Anything & everything by E.O. Wilson, the Pulitzer Prize-winning writer & scientist on ants, humanity, and social organisms.
Appendix: Slides & Video
This article is based on my keynote at the Startup Energy Transition (SET) Tech Festival in Berlin on April 16, 2018. Here are the slides, with the video embedded.
I gave a similar talk at Dezentral Blockchain conference in Berlin, Germany on Sept 8, 2018. In this case my emphasis was the future of work.
AI could take almost all the jobs, not just trucking and retail but even creative jobs. Can we address this in a way that actually puts humanity on a path to abundance? The Nature 2.0 approach might just do this, by leveraging the surplus of self-owning AI DAOs into Universal Basic Income (UBI), then, not stopping, into Universal Self-Actualization Income (USI).