The AI Anarchists Manifesto
One of the most powerful things about infinitely scalable server-less functions is that you can deploy everyday functions — and thus people’s jobs, and entire organisations of any size — as software (or, specifically, as cloud code).
I talked about this more in my article “Adios, Middleware Jobs“, but in a nutshell: Any job that’s just a series of inputs and outputs (e.g. take data and make a spreadsheet and a presentation, or take the opinion of the public and formulate it into policy and law) is at risk of becoming entirely redundant within the current generation.
By extension, many organisations will become entirely redundant. There are so many companies that are entirely back-end. Think of banks, for example. Take away branches — which is something banks desperately wish they could do — and they’d be entirely an army of people in an ivory tower doing trades, moving money here and there, and executing corporate strategy. Bam, the entire thing could vanish and be replaced by a series of functions.
const manageAccounts = new Workflow();
manageAccounts.addBusinessProcess({
"name": "signUpCustomer",
"inputs": CustomerDetails,
"processSteps":
...
});
There you go — that’s the starting point for a function that defines a lot of what people currently manually do. And it can easily scale up to be what people at the very top do.
const manageLegalSystem = new Workflow(countryName: String, population: People[]);
manageLegalSystem.actions.add({
"name": "createLegislation",
"inputs": [
{"name": "surveyData", type: "object"},
{"name": "legislation", type: "resource"}
]
"processSteps": [
"Analyse survey data to understand trends, preference, and current most pressing issues",
"Assess legislative framework to understand gaps relative to current population requirements",
"Draft legislation to fill legislative gaps",
"Get input from community management on contents, timing, and budget",
"Enact legislation"
],
...
}
That’s the starting point for a function that does what a lot of government does. There exists the potential for AI-driven functions to replace all our government with one that is completely completely, has no self interest, and is always working at essentially infinite scale.
People think of “anarchy” as being “chaos”. But in the fields of law and governance, anarchy is more literal — it means “without a leader” (see this Wikipedia article). That’s the kind of anarchy I’m talking about — destruction of an institution, yes, but the creation of the ultimate democracy. The ultimate democracy is one in which there is only the people.
Here’s the dangerous question: Given that many jobs, organisations, and maybe industries (call centres, BPO) will become obsolete, which ones should we actively seek to render obsolete?
In my opinion, but one that is shared by many, there are a number of cancerous private and public sector roles and organisations without which everyone would be better off. These include:
- Politicians and political parties. More broadly, government in any role that isn’t purely administrative. Many social problems start off because of inefficiencies, vested interests, incompetence and corruption here. Kill the snake by killing the head.
- Real estate agents. I can take my own photos and negotiate my own pricing. I don’t need to pay someone an arbitrary percentage to smile and lie.
- Headhunters / Recruiters. Modern day slave traders who do nothing more than scour LinkedIn and send form emails.
- Most financial services. Most banks exist as basically a vehicle for skimming a percentage out of everyone’s wealth. I’m including in this borrowing and lending, currency trading, investment services, brokerage services, insurance sales… the list goes on.
These all seem to be serving us in some way, but humans don’t exist without self interest. There wouldn’t be real estate agents if it weren’t profitable. There wouldn’t be politicians if people didn’t enjoy power.
The question now becomes twofold:
- If AI will create and destroy anyway, then should we actively seek to eliminate bad actors in society using AI? (Or really… Why shouldn’t we?)
- How can we use AI to eliminate these cancers from our society?
Should we eliminate industries using AI? If so, which ones?
The first question is: should we proactively eliminate cancerous concepts from our society? It seems like a stupid, almost rhetorical question.
To start with an easy-to-relate-to example, the majority of everyday people hate real estate agents. Poor people hate them for inflating prices and discriminating. The wealthy look down their noses at them. The only people who don’t hate real estate agents are other people who profit from the industry, like mortgage lenders, banks, builders, and governments.
Then to go all the way to the top, most people would laugh at the idea of an honest politician. I’m sure many aspire to be that. But party politics, vested interest, and “realism” get in the way of the ideals of many who aspire to be the politician that bucks the trend.
Then the question we must ask is: Which industries and social concepts should we prioritise in eliminating, and where should the reallocated wealth go?
If you ask me for my personal hitlist, I’d prioritise those that are publicly permitted and legal but which reduce public good.
- Government — It always shocks me to remember that our best and brightest usually don’t seek to work in government, whether in politics or in administrative positions. The bar is very low for entry and usually has little to do with how good a person (let alone how smart or hard working) you are.
- Real estate agents — They provide no value, and are incentivised to just produce a quick sale (maximising their own profits), pressuring buyers to pay more and sellers to accept less, thus being a key factor in why house prices rise.
- Loan brokers (home loans, car loans, etc.) — these very obviously can be replaced by a website or just an API
- Payday loan companies — Predatory and exploitative in the already troubled world of lending
- Headhunters / Recruiters — These are modern-day slave traders who hoard information, add no value, and just take a slice out of transactions that should be otherwise easy to facilitate
- Insurance brokers — Insurance barely has value anyway (they’ll take your money but kick up a fuss about paying out); there doesn’t need to be a disintermediating service
- Pharmaceutical marketers — These are largely responsible for inflated prices of often essential drugs.
- Credit rating agencies — Biased info and rife with conflicts of interest
- Utilities providers, e.g. telecommunications and energy, whose pricing can be predatory and confusing, and who provide an unnecessary layer between public infrastructure and the consumer.
- Private educational institutions — Often incentivised to maximise profits, which skews the nature of how they market their courses and the research they conduct.
Note that I’m not talking about actual criminal behaviour like scammers, drug sales, or illegal gambling, which absolutely should be targeted, too. (And I’m also not including my personal hitlist of things that I am against, like religious organisations.)
If a world existed without the above things, it’d be exactly the same as it is now — just much more stable and efficient. A world without politicians and real estate agents; could we really get there? And a world where no job requires an MBA? (I’m not really a fan…)
How can the above industries and organisations be eliminated using AI?
I’ve previously addressed the question of how it can be done broadly. But there should be a general extension to that concept of replacing “back-end” and “middleware” jobs to acknowledge the fact that there are many technologies that allow AI to replace the front-end, too.
For example, right now there already exist a number of real-time audio TTS and speech APIs. It will only be a matter of time and processing power until the same exists for video — and it does, in basic form, already.

To eliminate entire industries like real estate, brokerages, and even governments using AI, you would have to use LLMs to
- Communicate with people via a front end (live streaming, video generation, social media, surveys)
- Make decisions in the back end (this is by far the easiest part)
- Communicate via the front end or via third parties (e.g. to implement legislation or to enact parts of the budget)
As I discussed in “Adios, Middleware Jobs“, you can describe what most people do in terms of basic functions.
For example, a politician has to do these things:
- Gather needs from the electorate
- Know the local laws and constitution
- Understand party policy (assuming we haven’t eliminated parties)
- Know the budget allocations and forecast
- Make decisions to maximise public interest, minimise public harm, and most effectively spend the budget
- Communicate decisions with whoever needs to enact them
- Report on progress
- Review impacts and start again
Writing it out like that, it seems almost trivially easy to replicate using software. Part of the reason it seems easy is that I’ve actively ignored the worst parts of politics — campaigning, infighting, and damage control from scandals.
Here’s a basic spec for the above:
Front-End
- Communication Channels:
- Video calls: Use Twilio Video for video communication.
- Phone calls and SMS: Integrate with Twilio API for telephony and SMS.
- Email: Use Nodemailer for sending email reports and updates.
- Social media: Use Twitter API and Facebook Graph API for social media interactions.
- UI Framework:
- Next.js with React for the front-end interface.
- Real-Time Notifications:
- Socket.IO for real-time notifications and updates.
Data Gathering
- Input Sources:
- Surveys and forms via Google Forms API.
- Social media scraping using Twitter API and Facebook Graph API.
- Public records and policy documents via local government open data portals (custom API integration depending on the locality).
- Natural Language Processing:
- OpenAI API for sentiment analysis, summarization, and input processing.
- Compromise for lightweight NLP on simpler text inputs.
- Citizen Profiles:
- Store individual data in a structured format using any database (e.g. GraphQL + Prisma).
Decision-Making
- Core Logic:
- TensorFlow.js for predictive modelling and optimization based on budget constraints and public interest.
- OpenAI for LLM processing / decision making.
- Node-RED for building and integrating the rules-based decision-making process.
- Rules Engine:
- Use JSONLogic for declarative rules evaluation (to evaluate decisions based on law, policies, and public input).
Reporting
- Reporting Channels:
- Use React with D3.js for an interactive dashboard for decision-makers to view progress.
- Schedule email reports with Node Schedule and Nodemailer.
- Progress Tracking:
- Monitor impact metrics via real-time data visualization (D3.js) and analysis using Grafana for reporting.
- Feedback Loop:
- Regular review cycles with updates to decision models via automated retraining using TensorFlow.js and OpenAI.
Looking at the above, I immediately notice two things:
- I could replace the entirety of government with this one application.
- I only need one, not hundreds
So — lower cost, higher efficiency, and infinitely scalable. What’s not to like?
The Manifesto
Bearing the above in mind — that it’s both in the public interest to downsize society using AI, and that it’s easy and efficient to do so — it’s easy to write the following AI Anarchist’s Manifesto.
Using AI, we should:
- Eliminate Harmful Industries: Industries like real esate agencies, financial services, and for-profit healthcare exist primarily to profit from people’s needs, not to serve them. AI can replace or eliminate these, focusing on fairness and transparency, maximising efficiency, and minimising harm.
- Use AI for Public Good: Create new industries (or forms of existing ones) that are focused on human well-being and happiness, including governance, healthcare, and asset distribution to serve basic human needs.
- Recreate Democracy: Take the human element (with all its waste) away from government, and create automated governance systems that truly represent the electorate’s will, with no vested interests, politics, incompetence, or inefficiency.
There’s some overlap between the above, but it comes down to one thing — corruption, waste, and incompetence are costing our society, and they’re not features of automated systems.
Because I like the way it sounds, I’ll write it again: The ultimate democracy is one in which there is only the people. We can get there.