How AI is making human intelligence interoperable
For a long time, I thought intelligence was largely trapped inside the architecture of the individual mind.
You might have strong pattern-recognition ability, curiosity, imagination, analytical capacity, or an unusual ability to connect apparently unrelated ideas. But your output was constrained by the hardware and interfaces of your mind.
You could have the engine without having the transmission.
You could have the ideas without having the time to execute them.
You could understand a visual concept without being a particularly good visual designer. You could have mathematical intuitions without being trained as a mathematician. You could understand a programming problem without enjoying syntax, repetitive coding patterns, debugging, or memorizing programming conventions.
And therefore, society has often confused what a person can potentially think with what that person can practically produce.
I think AI is beginning to break that relationship.
1. The Mind Is Not the Whole System
Human intellectual output is not simply:
Intelligence → Output
It is closer to:
Intelligence × cognitive bandwidth × workspace × tools × time → output
A highly capable mind can still be bottlenecked by almost everything on the right side of that equation.
Time is finite. Attention is finite. Working memory is finite. Learning every specialized interface is expensive.
And much of what we do requires translating thought into mechanical execution. That translation can consume an extraordinary amount of prime cognitive real estate.
I know this from experience.
I have written around 3,000 articles. Writing itself isn’t my fundamental bottleneck. I can write.
My bottleneck is time.
I can also design things. I have designed logos and other visual material myself. But sometimes the process takes far too much of my cognitive time relative to the intellectual value I receive from doing it personally.
Why should I spend my best thinking hours manually producing something that can be meaningfully automated?
The question is not:
Can I do this?
It is:
Is this worthy of my cognition?
That is a very different question.
2. Intellectual Hacking
This is what I mean by intellectual hacking.
Intellectual hacking is not simply learning new skills. It is finding ways to bypass the bottlenecks that prevent your existing intelligence from being expressed.
AI provides an extraordinary new mechanism for doing this.
I may have an idea but limited visual-production ability. AI gives me a visual workspace.
I may have a software idea but dislike the conventional interface of programming. AI gives me another interface.
I may have an enormous number of ideas but insufficient time to turn all of them into polished articles. AI can take over much of the production machinery.
The important thing is that my underlying intellectual direction has not necessarily changed.
The interface has changed.
I have effectively hacked my way into capabilities that previously required much more specialized execution.
I can now move from:
thought → AI → artifact
instead of:
thought → years of skill acquisition → specialized execution → artifact
That is not merely a productivity improvement. It changes the boundary of what an individual can practically do.
3. AI as a Thought and Workspace Capacity Amplifier
I increasingly think we should stop describing AI merely as an intelligence amplifier.
That is incomplete.
AI is also a thought and workspace capacity amplifier.
AI doesn’t necessarily make my biological brain smarter. But it can give my brain more working space, more external memory, more rapid iteration, more alternative representations, more visualization, more implementation capacity, more conversational exploration, and more opportunities to test an idea.
In effect, the space in which my intelligence can operate becomes larger.
This is what I call soft brain extrapolation.
The biological brain remains the same. But its effective operating environment expands beyond the skull.
4. The Hidden Polymaths
This also makes me wonder whether we have systematically underestimated the number of potentially multidisciplinary people in the world.
There may be many people who are latent polymaths.
A good doctor may occasionally have an unusually good mathematical idea. An engineer may have a powerful philosophical insight. A mathematician may have a visual intuition. A writer may understand a software architecture surprisingly well.
But normally, the person encounters a wall.
They don’t possess the vocabulary, training, tools, credentials, or time required to cross into another domain.
So their intelligence remains domain trapped.
This is important because some apparent specialization may not reflect the actual boundaries of human intelligence. It may reflect the boundaries of access.
AI changes that.
5. Intellectual Interoperability
This leads to another concept I find increasingly interesting: intellectual interoperability.
Traditionally, specialization has meant that a person’s intelligence becomes connected to a particular domain.
A doctor connects intelligence to medicine. A programmer connects intelligence to software. An architect connects intelligence to buildings. A designer connects intelligence to visual communication.
AI potentially allows the same underlying cognitive engine to connect to multiple domain interfaces.
The person doesn’t have to become a full expert in every domain. They need enough understanding to direct, interrogate, evaluate and integrate the work.
Much of STEM work, at a sufficiently abstract level, involves some combination of pattern recognition + domain knowledge + tools + procedural competence + judgment. The proportions differ enormously between fields.
AI increasingly supplies portions of the domain interface and procedural machinery.
That leaves more room for the human to ask:
- What is the pattern?
- What is the problem?
- What doesn’t fit?
- What follows from this?
- What should we investigate next?
Take my own raw intelligence and pattern-recognition ability and put it somewhere else.
That is becoming increasingly possible.
6. Vibe Coding and the Interface Problem
This is why I don’t dismiss vibe coding as merely a temporary curiosity.
The conventional model of programming assumes that someone who wants to create software must first become proficient in the programming interface.
But perhaps that assumption was partly an artifact of the technology.
Programming languages require syntax. They require memorization. They contain repetitive patterns. They involve debugging. They require a particular style of procedural thinking.
None of those things necessarily constitutes the essence of software problem-solving.
A highly intelligent person may understand systems, algorithms and patterns very well while simply having little interest in the traditional programming interface.
That was partly my own experience. I even let go of pursuing an MSc in IT.
It wasn’t necessarily because I lacked the intellectual capacity to understand computing. The interface itself was not attractive enough to justify occupying my cognitive bandwidth.
AI changes that interface.
Now I can potentially express a software idea in natural language, inspect what is produced, reason about it, modify it and iterate.
I don’t have to become a traditional programmer before I can participate in software creation.
That is intellectual hacking.
7. The AI Doesn’t Make Everyone Equal
But there is another misconception hiding inside the AI-assisted coding debate.
If everyone has access to essentially the same AI, will everyone produce roughly the same quality?
No.
The important advantage may increasingly lie in knowing what to ask for, what problem to solve, what constraints matter, what to reject, and what to try next.
This is hardly unique to AI.
Same books. Same teacher. Same school. Different results.
Education has never produced identical outcomes simply because people received access to the same information. The difference is in the minds receiving and using that information.
The same principle applies to AI.
Two people can have access to the same model and ask it to build the same application. One may produce something mediocre while another discovers a better architecture, identifies hidden edge cases, asks better questions, iterates intelligently and recognizes possibilities the first person never considered.
The AI may be similar.
The intellectual direction is not.
This is precisely why AI can become an amplifier rather than merely an equalizer.
A weak idea amplified can remain weak.
A strong pattern-recognizer with an enormous AI workspace can become extraordinarily productive.
The scarce resource may gradually move from the ability to execute instructions to the ability to formulate the right instructions in the first place.
And even “knowing what to ask” is only the surface.
The deeper abilities are:
knowing what is worth asking → recognizing an interesting answer → seeing what is missing → asking the next question → connecting the result to something else.
That is scientist-like thinking.
And it is also intellectual hacking.
8. The Compression of Specialization
This has a potentially enormous economic consequence.
Suppose one person can use AI for writing, programming, research, visualization, analysis and design.
That person doesn’t necessarily possess six professions. Instead, their intelligence has become interoperable with six professional workspaces.
Historically, specialization was partly necessary because the cost of acquiring and operating those interfaces was high.
AI lowers that cost.
The consequence may be a compression of specialization.
One highly capable person may increasingly perform work that previously required several specialists.
That does not necessarily mean that every specialist disappears. But it could mean that fewer people are required to produce the same amount of intellectual output.
This is where AI-induced joblessness becomes more interesting than the usual “robots will take our jobs” story.
The disruption may not require AI to become universally autonomous.
It may simply require:
One capable human + AI to cover the productive surface previously covered by several humans.
And then comes the uncomfortable question:
What happens to everyone else?
A lot of peaceful people may simply be waiting outside for the peaceful division of the wages created by dramatically increased productivity.
The machines don’t have to revolt. The humans don’t have to revolt.
The economic problem can arise simply because the arithmetic changes.
9. From Jobs to Cognitive Capacity
Perhaps we have been organizing the economy around the wrong unit.
The traditional model is roughly:
person → skills → occupation → job
The emerging model may look more like:
person → cognitive capacity → AI workspace → output
That is a profound shift.
The value of an individual may increasingly depend not only on what specialized skills they have accumulated, but on how effectively they can connect their intelligence to different tools and domains.
This may create a new kind of person:
Not necessarily a polymath.
Not necessarily a specialist.
But an intellectually interoperable human.
Someone who can take the same underlying intelligence and deploy it in different environments.
10. And Perhaps Science Is Coming Back
There is an interesting historical irony here.
For a long time, economic systems rewarded enormous amounts of execution.
Learn the process. Follow the procedure. Master the tool. Produce the output.
But when machines become increasingly good at execution, the relative value of figuring out what should be done may rise.
That brings us back to something resembling scientific thinking.
Observe. Question. Connect. Hypothesize. Experiment. Inspect. Revise.
AI can increasingly handle large portions of the workspace surrounding this loop.
The human can spend more time deciding what is worth thinking about.
So perhaps we are entering a period in which society begins to reward scientist-like thinking more broadly—not necessarily by employing everyone as a scientist, but by making scientific modes of thought economically useful across many domains.
I think time may be returning to reward scientist-like thinking.
Not because execution has become irrelevant, but because increasingly capable machines are taking over more of the execution.
The scarce resource becomes good questions.
And perhaps even more importantly:
the mind capable of recognizing which questions are worth asking.
11. A New Paradigm
This may be the beginning of a new paradigm.
For centuries, human intelligence was constrained by the cost of turning thought into action.
AI is beginning to collapse that cost.
It doesn’t necessarily give everyone the same intelligence. It gives more people access to increasingly powerful interfaces for expressing whatever intelligence they already possess.
That distinction is crucial.
The future may therefore contain more people who can move between domains without fully becoming specialists in each one.
More people may discover capabilities they didn’t know they possessed.
More people may become latent polymaths made operational.
And highly capable people may become extraordinarily productive because their prime cognitive real estate is no longer consumed by low-value execution.
This is soft brain extrapolation.
This is intellectual hacking.
This is intellectual interoperability.
And perhaps this is the most important shift of all:
AI may not give us a new brain. It may give our existing brain more places to go.
The biological brain remains the engine.
AI expands the workspace.
Intellectual hacking chooses the route.
Intellectual interoperability determines where the intelligence can go.
And perhaps, for the first time, a person’s potential may be constrained a little less by what they were trained to be and a little more by what they can imagine doing with the intelligence they already possess.