← All writing Draft · Nov 2025

The Rise of the Outcome Economy

The near-term economic transformation. Knowledge workers become intelligence workers, and outcome-based economies emerge.

Exploring the Near-term Impacts of AI on Knowledge Workers

Buddy Williams · November 7, 2025

Knowledge advances through bold conjectures and rigorous attempts to falsify them. —Inspired by Karl Popper

A Common Question

When people read Technohumanism, often their first question is:

“What will happen to jobs?”

It’s an understandable question. AI feels both thrilling and threatening—a force poised to reshape everything. My goal in writing Technohumanism was to explore humanity’s relationship with technology more broadly, but the jobs question lingers because it’s deeply personal.

Like you, I’ve heard both extremes:

“AI will end all jobs.”
“AI will create more jobs than it destroys.”

Who’s right?

This essay builds a model to think clearly about that question. The test of any good model is whether it helps us navigate reality. So, let’s explore this together.

AI → Work → Impacts → Economy → Abundance

About Me

You might be wondering who I am and why you should believe what I have to say. That’s an entirely fair question. Pursuing truth is my passion. Truth can come from anyone. So, let’s frame this as two agreeable strangers thinking together.

I’m a leader with 25 years of experience in software development. I lead AI initiatives that span research, product delivery, and strategy.

In the end, I don’t think you should listen to me because of who I am or what I do; judge for yourself whether what I have to say is worthwhile. Primarily, I want to encourage you to think for yourself. Let’s begin.

Nature of AI

Before we can reason about jobs, we have to understand what AI actually is. We do this by making observations. This is the first step of reason. I try my best to see the nature of the thing I’m observing. What are its fundamental properties? What makes it unique or distinct?

Large language models like ChatGPT are prediction engines — they predict the next word or token based on probabilities derived from vast patterns of data.
They are lossy compressions of collective human knowledge: not perfect memories, but flexible reconstructions of patterns.
They can generalize, generate, and reason across domains.
They increasingly use tools, planning, and multi-step reasoning over long time horizons.

Understanding what AI is reveals what it naturally tends to automate.

Capabilities

Today, AI systems can:

Generate, translate, and classify text, images, audio, and video.
Explain, reason, and converse naturally.
Use tools, browse data, and perform research.
Write, test, and debug complex software.
Support scientific inquiry and design novel materials.

Summed simply: AI automates thinking. Where the Industrial Revolution automated muscle, AI automates knowledge work.

AI is a thought calculator — automated intelligence.

Three Worker Classes

Jobs can be roughly divided into three economic categories:

Field Workers – Physical labor (construction, repair, logistics).
Knowledge Workers – Intellectual labor (medicine, law, engineering, software).
Asset Workers – Capital labor (finance, investing, ownership).

AI touches all three, but the knowledge worker is most directly impacted. The expertise moat is draining fast. AI can already perform most forms of informational labor and will soon do so faster, cheaper, and better.

Low-level Impacts

Comparing before AI and after shows what might change, specifically, specialization and implementation. These flow directly out of the AI capabilities above: knowledge, reason, and implementation.

Deflation of Specialization

Before: If you wanted to write computer software, you’d need days of study to do the most basic tasks and years to perform complex assignments.

After: We ask AI to write code for us, explain it to us, and compare it with other possible approaches — no programming expertise required. It is a calculator for thinking. A tangential scenario plays out across all knowledge work — no specific knowledge is needed.

Deflation of Implementation

Before: If you wanted to build a complex software system, you’d need a team of developers, architects, and coordinators to perform manual implementation. Every field has time spent searching and deriving the correct information in the proper context.

After: Spin up a number of agents with higher-quality capabilities who will work 24/7 on implementation. Agents will speed up every field. In my field, software development, we are already seeing this implementation speed up.

The Rise of the Intelligence Worker

AI automates knowledge work, but it doesn’t make humans obsolete. Knowledge workers become intelligence workers — collaborators in a new creative discipline that extends human thought through AI. Their craft centers on three primary functions:

Bold Conjectures — generating ideas that push boundaries.
Orchestration — managing AI workers.
Evaluation — interpreting what’s valuable, true, and meaningful.

Where knowledge work applied existing understanding, intelligence work explores the unknown. It’s the shift from execution to exploration.

The outcome is an Idea Frontier Expansion — a global acceleration of discovery as millions of humans and AIs co-create, each building on the other’s output.

Falsification

Science must begin with myths, and with the criticism of myths.

Above, I’ve argued that knowledge and implementation are being commoditized, transforming knowledge workers into intelligence workers, leading to outcome economies. I’ve only presented logical arguments, no evidence. I’ve hoped you would suspend your inner critic. This is entirely on purpose. People are addicted to data. Actors weaponize data to justify their positions. At the same time, people are not working through hypotheticals. Our reasoning habits have been bifurcated. When it comes to data, we really want a range of tools to explore truth.

All models should invite challenge. This one fails if any of the following prove true:

AI does not deflate knowledge work. Data showing that while AI can provide knowledge, it cannot do so satisfactorily, so the need for specialized knowledge workers will continue.
AI does not deflate implementation effort. Data showing that while AI can do some tasks, a human-in-the-loop will persist.
Reallocation to outcomes-style businesses fails. You could show that reallocation will flow more naturally into field and asset work, or even some new form of work.
Society rejects or limits AI deployment through regulation or collapse.

Let’s test this model against emerging signals.

Evidence

Signals supporting this transformation already appear:

ChatGPT, with hundreds of millions of users, has normalized thinking with machines. Roughly half of interactions are exploratory (“asking”), not just productive (“doing”).
AI-led education models like Alpha School and Khan World School demonstrate structured, individualized instruction at scale — bespoke intelligence amplification.
Autonomous software agents like Blitzy, Claude Code, and Codex show implementation deflation: complex products built in days, not months.
AI Scientist agents, such as Kosmos and AlphaEvolve, are emerging with novel discoveries.
GDPval benchmarks reveal that frontier models now perform near-expert levels across dozens of knowledge industries.
McKinsey and MIT research show AI doubling innovation rates in R&D-driven sectors — the measurable birth of the Frontier Idea Explosion.

These signals point in one direction: rapid transformation.

Economics of Abundance

An economy is a system for organizing effort toward desired outcomes. When the cost of effort collapses, the system’s structure must change. In the Industrial Age, value came from machines amplifying muscle. In the Information Age, it came from computers amplifying knowledge. In the Outcome Age, it comes from intelligence amplifying creativity.

Companies will shrink in size but grow in both numbers and capability. Their competitive edge will no longer be manpower, but clarity of mission. Most organizations will look like small mission pods — humans and AIs aligned around a single objective, coordinating through shared intelligence networks.

At the macro level, three results emerge:

Capital Concentration — Owners of productive AI systems initially capture most value, and wealth inequality skyrockets.
Redistribution — such as compute dividends and universal income through taxes. This is inevitable for two reasons:
- Businesses need consumers, and consumers need capital.
- People have a desire to have collective dignity.
Outcome Economies — Capital flows to exploring the idea frontier of abundance progress. Success is measured by solved problems and a more abundant world. Money becomes a tool for directing intelligence; the scarce resource becomes human meaning.

The Outcome Economy

As AI systems automate the “how,” the value of implementation collapses — but the value of direction explodes. What matters is no longer the ability to produce, but the ability to define, explore, and evaluate outcomes worth pursuing.

Intelligence workers drive the Outcome Economy — a system organized not around implementation, but around goals.

They convert intelligence into progress across three frontiers of abundance:

Biological — The domain of health, longevity, and genetics. The resulting abundance is time, more life to live.
Material — The domain of energy, robotics, and materials. The resulting abundance is more goods and physical prosperity.
Social — The domain of play, competition, and connection. The resulting abundance is meaning, fun, richer lives, morality, and connection.

When knowledge is free and implementation is instant, the scarce resource becomes insight. Intelligence workers supply it — framing questions, steering exploration, and synthesizing value. The old economy was built on labor power. The new one is built on intelligence power.

We are moving from economies of effort to economies of direction — from managing people to managing intelligence.

A New Identity

Imagine a world where no one asks, ‘What do you do for a living?’ but instead, ‘What outcome are you working toward?’ This is where AI is taking us, to outcome economies. Many will enjoy their work more because of greater control over where to find meaning.

As the barrier for knowledge work vanishes, the generalist will be empowered to leverage any specialty field by having an AI expert work alongside them. The future will belong to quests, missions, and goals. There will be an explosion of accomplishments as smaller groups of people organize to pursue causes. You won’t be a graphic designer. You’ll be a cancer-cure maker.

I’ve heard it said that the future of work belongs to entrepreneurship. This is not far from the truth. Instead of focusing on building a business, the focus will be on causes. We see hints of this in The Network State, where people organize around a moral cause. John, a contributor to the popular rationality website LessWrong, describes this as Orienting Toward Wizard Power.

The transition to outcome economies won’t be frictionless. People need new forms of economic dignity. Some form of universal income or compute dividends. The challenge ahead isn’t whether AI can create abundance, but whether we can redistribute ethically.

We may be wrong, and even the best-tested theory is only a conjecture.

Advice for Navigating the Transition

1. Use AI to Think Better

Don’t just use AI to produce output; use it to expand understanding. Collaborate with it as a partner in reasoning, reflection, and creativity.

2. Think in Outcomes, Not Jobs

If your income depends on hours billed or narrow expertise, you’re standing on melting ice. Explore ideas you find interesting, and let AI implement them.

3. Invest in Coordination and Meaning

When intelligence is abundant, purpose is scarce. The rarest and most valuable skill will be aligning humans around shared goals.

4. Build a Nest Egg

Keep your fixed costs low and be flexible. In a volatile world, adaptability is wealth.

5. Cultivate Falsifiability

Treat every belief as a working hypothesis. Knowing how your worldview could be wrong is the surest way to stay adaptive.

Conclusion: From Knowledge to Outcome

The future is uncertain, but I believe we are witnessing the next major transition in the story of work: Knowledge → Intelligence.

Knowledge workers evolve into intelligence workers — people who generate conjectures, orchestrate AI workers, and evaluate meaning. Their work fuels a global idea explosion, leading to biological, material, and social abundance.

In outcome economies, wealth flows to those who direct intelligence toward meaningful ends. This is not the end of work — it’s the beginning of intelligence work. And it’s the dawn of an age where the question is no longer “What do you do?” but “What outcome are you helping to create?”

Appendices

Appendix A: Tracking AI Progress

Appendix B: Is AI an Effective Teacher?

By default, large language models offer unstructured learning: you ask a question, get an answer, and move on. Used alone, this approach risks shallow understanding — leading to traps like focalism, Maslow’s hammer, or centrality bias. Unstructured learning can produce confidence without comprehension — the illusion of understanding.

Fortunately, AI is already moving beyond this. Systems like Alpha School and Khan World School show how AI can deliver structured, personalized learning — meeting students where they are and guiding them to where they need to be.

This shift enables bespoke education: “just-enough” and “just-in-time” learning tailored to the individual. It empowers generalists to compete with specialists by outsourcing detailed expertise to AI while focusing on creative reasoning and direction.

In short, AI won’t make us dumber; it can make learning adaptive. It replaces rote accumulation with continuous, context-aware understanding — freeing humans to spend less time memorizing and more time thinking.

Sample AI Education Programs:

Alpha School — private K-12 network where students do core academics in ~2 hours via AI/adaptive software
Unbound Academy — model explicitly prioritizing AI to deliver core academics
Khan World School @ ASU Prep — online high school integrating Khanmigo AI tutoring
2-Hour Learning — AI-tutor framework marketed as “crush academics in 2 hours”
Super Teacher — low-cost AI tutor for elementary schools
Khanmigo (Khan Academy) — AI tutor & teacher assistant deployed across districts
Kira Learning — AI co-teacher agents for one-on-one guidance
Kyron Learning — AI-driven interactive lessons and conversational video tutors
Knewton Alta (Wiley) — adaptive courseware with AI-personalized pacing
Alice.Tech — “Duolingo for exams” with AI-generated personalized study plans
MagicSchool — widely adopted AI platform with student tools and tutors