Intelligence Has Always Been Expensive
For most of human history, intelligence has been scarce and costly.
Training a doctor takes decades. Writing software requires years of practice. Strategic thinking, creativity, and deep domain expertise have traditionally been locked behind time, education, and human attention. Even with computers, intelligence remained a human bottleneck: machines could compute, but people had to think.
That constraint shaped entire economies. High-skill labour commanded high wages precisely because intelligence could not be cheaply replicated.
AI is breaking that assumption.
AI as a Cost-Collapsing Technology
Artificial intelligence is not just a new tool it is a cost-reduction engine for cognition itself.
Each generation of models:
- Requires less human input per unit of useful output
- Performs more tasks once considered “expert-only”
- Scales almost infinitely at near-zero marginal cost
Once trained, an AI system can:
- Write code
- Draft legal documents
- Diagnose medical images
- Tutor students
- Design systems
- Generate art and ideas
And it can do so millions of times in parallel.
This mirrors what happened with:
- Electricity replacing manual labor
- Calculators replacing arithmetic
- Compilers replacing assembly
- Cloud computing replacing physical servers
Except this time, the thing being automated is thinking.
The Declining Marginal Cost of Intelligence
The most important economic shift AI introduces is this:
The marginal cost of intelligence is approaching zero.
Once a model exists:
- Serving another user costs fractions of a cent
- Producing another answer is effectively free
- Replicating “expertise” is instant
This is radically different from human intelligence, where:
- Each additional expert requires years of training
- Burnout and attention are real constraints
- Scaling is linear and slow
AI intelligence, by contrast, is:
- Non-rivalrous
- Infinitely copyable
- Improving faster than its deployment cost
This is the same curve we’ve seen before just applied to cognition instead of computation.
When Intelligence Becomes Infrastructure
As costs fall, intelligence stops being a product and becomes infrastructure.
We no longer think about:
- The cost of arithmetic
- The cost of storing data
- The cost of transmitting information
Similarly, we are moving toward a world where we no longer think about:
- The cost of drafting text
- The cost of basic reasoning
- The cost of first-pass problem-solving
Intelligence becomes:
- Always available
- Embedded everywhere
- Assumed by default
Just as software ate the world, AI is eating mental overhead.
Zero Cost Doesn’t Mean Zero Value
A common mistake is to assume that if intelligence becomes cheap, it becomes worthless.
The opposite is usually true.
When a constraint disappears:
- New products emerge
- Expectations rise
- Leverage shifts to those who ask better questions
When computation became cheap, software exploded. When information became free, curation mattered more. When intelligence becomes abundant, judgment, taste, and direction become the scarce resources.
The value moves up the stack.
What Still Costs Something
Even in a near-zero-cost intelligence world, some things remain expensive:
- Problem selection – knowing what to work on
- Context – understanding real-world constraints
- Accountability – owning outcomes
- Taste – recognizing what “good” looks like
- Trust – deciding when and how to rely on systems
AI can generate answers, but humans still decide which answers matter.
At least for now.
The Asymptote: Effectively Zero
The cost of intelligence may never be mathematically zero but it doesn’t need to be.
Just as:
- Storage is “free enough”
- Bandwidth is “free enough”
- Compute is “free enough”
AI-driven intelligence will become cheap enough to be assumed.
When that happens:
- Not using AI will feel like not using a calculator
- Manual cognition will be a choice, not a necessity
- Entire workflows will be redesigned around this assumption
The real transition isn’t technical, it’s cultural.
The Question That Matters
The most important question isn’t if intelligence becomes free.
It’s:
What do we build when thinking is no longer the bottleneck?
History suggests the answer will be: things we haven’t imagined yet because imagining them used to be too expensive.
Now it isn’t.