Sequel to How AI Is Making the Lazy Brain More Productive
In the previous article, I looked at a simple but increasingly important phenomenon: AI can reduce the friction between having an idea and doing something with it.
A person may have an article in mind, a business idea, a mathematical question, or simply something they want to understand. The traditional bottleneck was often execution. AI removes much of that friction.
But there is another problem.
For some people, the difficulty is not getting ideas out of the head.
It is keeping track of where those ideas go.
Ideas do not always develop in a straight line.
They jump.
A question about mathematics can trigger a thought about artificial intelligence. That thought can connect to markets. A market observation can unexpectedly resemble something from psychology. Weeks later, the psychology idea can return while working on an entirely different mathematical problem.
The connections may look random from the outside.
But they may not be random at all.
They may be part of the way the person thinks.
Ideas Don’t Live in Folders
Most of our information systems assume that ideas belong somewhere.
Mathematics goes in one folder.
Writing goes in another.
Finance gets its own notes.
AI gets another.
This is convenient for storage.
It may be terrible for discovery.
The important connection between two ideas is often precisely the fact that they do not belong to the same folder.
A conventional notebook asks:
“Where should I put this idea?”
A more useful system might ask:
“What other ideas could this connect to?”
That is a fundamentally different question.
And AI creates the possibility of building such a system because it can work across large amounts of accumulated material while retaining relationships between pieces of information.
But before going further, there is an important distinction.
What Is Observed — and What Is Proposed?
The observations are relatively simple.
Some people generate ideas across multiple domains rather than developing one subject at a time.
Their thought process can be interleaved.
One line of thought may remain unfinished while another develops. An old idea may return much later. New information acquired in the meantime may change the meaning or usefulness of the old idea.
The same person may repeatedly move between apparently unrelated subjects.
These are observations about a pattern of thinking.
The Fractal Brain is something different.
It is a proposed model for representing that pattern.
There is no claim here that the human brain is literally a mathematical fractal.
The proposal is that a fractal-like network may be a useful representation of how interconnected ideas develop over time.
That distinction matters.
The model is a hypothesis about representation, not a biological claim.
From Linear Notes to Networks
Imagine that every significant idea is represented by a node.
Related ideas acquire connections.
Strong recurring ideas become hubs.
Several related ideas form a cluster.
Clusters connect to other clusters.
Eventually, larger structures emerge.
At the smallest level:
idea → related idea
At the next level:
idea cluster → idea cluster
At a larger level:
network → network
And the same relationship can appear repeatedly at different scales.
That is where the word fractal becomes useful.
The individual idea has structure.
The cluster has structure.
The network has structure.
The larger intellectual ecosystem has structure.
The claim is not that these structures are mathematically identical.
The useful analogy is that similar patterns of connection can recur at different scales.
The Cross-Boundary Jump
One of the most interesting features of this model is the cross-boundary connection.
Suppose someone is thinking about a problem in mathematics.
Later, while reading about markets, they notice a structural similarity.
The market observation changes the way they think about the mathematical problem.
Nothing about the two subjects requires them to be connected.
The connection comes from the thinker.
This kind of jump is difficult for conventional knowledge-management systems.
A folder system sees two unrelated subjects.
A network sees a possible bridge.
The bridge may eventually prove useless.
That is fine.
Creative thinking requires the ability to form connections that sometimes fail.
A system that only preserves validated connections may be excellent at documentation but poor at discovery.
Interleaved Idea Generation
There is another important pattern.
Ideas do not necessarily arrive one at a time.
Several intellectual streams can run simultaneously.
An unfinished mathematical question may remain in the background while the person works on an article.
An observation from the article may trigger a thought about AI.
That thought may lead to a completely different question.
The original mathematical problem has not disappeared.
It has simply moved into the background.
This can be called interleaved idea generation.
The mind does not necessarily complete:
A → finish A → B → finish B → C
It may operate more like:
A₁ → B₁ → C₁ → A₂ → D₁ → B₂ → A₃ → C₂
The sequence looks chaotic when viewed chronologically.
The underlying network may be much more coherent.
That distinction is important.
The Spaced Idea Box
Time introduces another dimension.
Some ideas should not be forced to mature immediately.
An idea can be parked.
It can remain unresolved.
Days, weeks, or months later, something new may provide the missing connection.
This creates what might be called a spaced idea box.
It is not simply an archive.
An archive says:
“Keep this because it may be useful later.”
A spaced idea box says:
“Keep this unfinished because future information may change what this means.”
The difference is subtle but important.
Time becomes part of the creative process.
The gap between two thoughts is not necessarily wasted time.
It can be incubation.
Return and Recombination
The most interesting event may occur when an old idea returns.
It does not necessarily return unchanged.
The thinker has changed.
The surrounding knowledge has changed.
The problem has changed.
Therefore the old idea can acquire a new connection.
Imagine an idea created on Day 1.
On Day 20, a completely different concept appears.
On Day 40, the two connect.
The resulting idea may be something that neither original thought contained by itself.
This is recombination.
The breakthrough is not necessarily the creation of a completely new primitive idea.
It may be the discovery of a relationship between ideas that previously existed separately.
Why AI Changes the Equation
This is where AI becomes particularly interesting.
Human memory is powerful but selective.
Chat histories, notebooks, bookmarks, documents, screenshots and scattered files are powerful but fragmented.
AI can potentially sit between them.
It can retrieve old ideas.
Compare them.
Notice recurring concepts.
Find connections across domains.
Track how an idea evolved.
And, perhaps most importantly, help preserve unfinished ideas without forcing them into premature conclusions.
That creates a possible division of labour.
The human generates.
The human makes unexpected connections.
The human decides what matters.
AI helps capture, retrieve, compare and map the growing network.
The AI does not have to become the source of every idea.
It can become the external memory and connection layer around the thinker.
From Chaos to Structure
There is a danger here.
If AI is asked to “organize everything,” it may destroy exactly what makes the system valuable.
It may turn a network into folders.
It may convert hypotheses into facts.
It may remove contradictions.
It may collapse several stages of an evolving idea into one polished paragraph.
The result will look cleaner.
It may also be intellectually poorer.
The objective should therefore not be:
Clean up the chaos.
It should be:
Find the structure inside the chaos without destroying the connections that produced it.
That is a very different task.
The system should preserve uncertainty.
It should distinguish an observation from an interpretation.
A hypothesis from an established result.
An abandoned direction from a dead idea.
A recurring connection from a one-time association.
And, crucially, it should preserve the history of important transformations.
The AI as Cartographer
This suggests a different metaphor for AI.
Not secretary.
Not search engine.
Not merely assistant.
Cartographer.
The thinker explores.
The AI maps.
The map does not determine where the explorer must go.
It records where the explorer has been, identifies possible bridges, and makes distant regions easier to reconnect.
Over time, the map itself becomes valuable.
It can reveal that apparently unrelated projects repeatedly use the same underlying concepts.
It can show that one idea has influenced five different areas.
It can reveal abandoned branches that may now deserve another look.
It can show where genuine intellectual hubs have formed.
This is potentially much more valuable than simply having thousands of pages of searchable notes.
The Fractal Property
The proposed model becomes especially interesting when viewed at different scales.
At the micro level, one idea connects to several related ideas.
At the meso level, those ideas form a cluster.
At the macro level, clusters connect into an intellectual network.
At an even larger level, several networks may interact.
The pattern repeats.
An idea can therefore be understood both as an individual object and as part of a larger structure.
This is why a network representation may be more appropriate than a list.
A list preserves items.
A network preserves relationships.
And relationships are often where the intellectual value lies.
From Productivity to Cognitive Infrastructure
This takes us beyond the original “lazy brain” argument.
The first transformation was about reducing friction.
AI made it easier to turn intention into action.
The next transformation may be about reducing cognitive loss.
Ideas that previously disappeared into old notebooks, forgotten chats, unfinished drafts or memory may remain connected.
The system does not merely remember that an idea existed.
It remembers its neighbours.
That could fundamentally change long-term intellectual work.
Instead of repeatedly starting from zero, a person could gradually build an external representation of their own evolving thought.
Not a database of everything they have said.
A map of how their ideas connect.
The Big Question
This model raises a larger question.
What happens when a human thinker has an AI system that has spent years observing the evolution of that person’s ideas?
Not merely their preferences.
Not merely their search history.
But their hypotheses, questions, abandoned paths, recurring intuitions, intellectual jumps, contradictions and breakthroughs.
The result could be something qualitatively different from ordinary personalization.
It could become a personal intellectual map.
The AI might eventually recognize:
“This new idea resembles something you explored two years ago.”
Or:
“These two projects appear unrelated, but both depend on the same underlying concept.”
Or:
“You abandoned this direction because of a problem that may no longer apply.”
That is not simply productivity.
It is cognitive continuity.
The Fractal Brain
The phrase Fractal Brain should therefore be understood as a model, not a diagnosis.
The idea is simple:
Thinking may be better represented as an evolving network than as a linear sequence.
Ideas branch.
Ideas cross boundaries.
Ideas wait.
Ideas return.
Ideas recombine.
Clusters form.
Larger structures emerge.
And the process repeats.
AI may eventually make it possible to preserve that process rather than forcing it into the linear systems designed for ordinary information storage.
The ultimate productivity gain may therefore not be producing more documents.
It may be losing fewer connections.
A human being generates the sparks.
Time allows some of them to mature.
Connections allow them to collide.
AI can help keep the map.
And when the right ideas finally meet, something new can emerge.
Ideas × Time × Connection = Discovery.