On intelligence

Intellect is among the most consequential human differences.

We accept variation in height, speed, coordination, musical ability, patience, endurance, and physical strength without much hesitation. We are less willing to accept variation in how easily people learn, reason, and manage complexity.

We treat cognition as equivalent to moral worth. It is not.

Aptitude is not character. It does not measure kindness, courage, judgment, taste, discipline, or the value of a life. But it is real, unevenly distributed, and consequential. It shapes how readily people learn unfamiliar systems, hold competing ideas in mind, recognize patterns, and respond when a task changes before they have mastered the previous version.

The useful question is not whether differences in intelligence are permissible to notice. The more important question is what schools, employers, and AI systems should do with the fact that those differences exist.

What IQ measures

IQ is not a complete account of the mind. It is a score produced by standardized tasks: verbal comprehension, reasoning through unfamiliar problems, working memory, processing speed, spatial reasoning, and quantitative thinking.

The reason these tasks can be combined into one broad measure is straightforward. People who perform well on one demanding cognitive task tend, on average, to perform well on others that are unrelated. This pattern is called general intelligence, or g. It does not explain everything a person can do, but it captures something stable:

Some people learn and reason through complex, unfamiliar material more easily than others.

This matters most when the rules are unclear.

A routine can be learned, followed, and improved by many different kinds of people. But when several variables must be held together, when the answer is not in the manual, when conditions change quickly, or when a plausible answer may still be wrong, differences in cognitive ability become more visible.

Intellect is not worth. It is capacity for cognitive complexity.

Inheritance and environment

Intellectual ability is partly inherited. The best evidence suggests that genetic influence on general cognitive ability is substantial and tends to increase as people get older, while the effect of the shared home environment tends to fade.

This does not mean that a person’s future is fixed at birth. It means that effort, parenting, and schooling are not complete explanations for why people differ in learning speed or tolerance for abstraction.

Environment matters. Improvements in nutrition, health, education, and the cognitive demands of modern life were associated with major increases in average test performance across much of the twentieth century. Later reversals in several developed countries also appear to have environmental causes.

Genes shape a range. Environment influences where someone lands within it.

Neither eliminates the other.

Intellect and discipline

One of the most useful distinctions in this subject is also one of the most neglected.

Intelligence and conscientiousness are different traits.

A person can learn quickly and still be unreliable. They can understand a difficult system and lack the discipline to finish anything inside it. Another person can be ordinary in abstract reasoning but organized, dependable, and exceptionally effective at work that rewards consistency.

Research finds little meaningful relationship between general cognitive ability and conscientiousness. They are not two measures of the same hidden quality.

They also matter in different situations.

Intellect helps most when a person is learning a new task, solving an unfamiliar problem, or adapting to changing conditions. Conscientiousness matters most when a task is already understood and needs to be performed carefully, repeatedly, and without excuses.

Neither trait is enough by itself.

High ability without follow-through often becomes unrealized potential. Discipline without enough capacity for the task can become a long struggle against a ceiling that effort alone cannot remove. The best outcomes usually come from a workable combination of both: enough ability for the work’s demands, and enough discipline to make that ability useful.

Complexity and fit

Job titles often conceal the thing that matters most: complexity.

Some work is stable. The rules are clear, the process repeats, and competence is largely a matter of training, attention, and reliability. Other work remains new even after years of experience. The variables shift. The information is incomplete. The cost of a bad decision can spread beyond the person making it.

Cognitive ability becomes more predictive as work becomes more complex; it matters most in work that requires continual learning, abstraction, and judgment under uncertainty.

Career titles obscure the underlying requirement. The more useful question is how much sustained novelty, abstraction, and independent judgment the work demands. The ranges below are approximate guides to the cognitive complexity a role typically demands—not career assignments. A serious matching system would use them as an initial filter, then consider conscientiousness, openness, interests, values, practical skills, and real constraints.

The ranges overlap because jobs overlap. A title is not a person, and an occupation is not a fixed task. The table is useful only if it is treated as a rough map of cognitive demand, not a deterministic catalog of who belongs where.

That does not divide work into “important” and “unimportant.” It describes different demands.

A skilled tradesperson may need spatial judgment, fine motor skill, practical intelligence, physical coordination, and a tolerance for real-world consequence that an office worker never encounters. A nurse, chef, technician, sales leader, mechanic, teacher, founder, engineer, and manager can all be highly capable in different ways. The relevant question is not whether one role has more status. It is whether the demands of the role fit the person doing it.

This is why the four-year college degree has become such an inadequate default. It treats one form of abstraction as the central path to adult competence, then leaves too few respected alternatives for people whose strengths are technical, practical, interpersonal, entrepreneurial, or physical.

The goal should be matching, not sorting.

A good system gives people multiple routes into useful, respected work. It does not funnel everyone through the same credential, then mistake the people who do not fit it for failures. A fuller matching system would account for more than cognitive ability: it would also consider discipline, openness, interests, values, practical skills, and constraints. That is the subject of On psychometrics.

What intellect is worth

Intelligence correlates with income, but not nearly as cleanly as popular arguments imply. The relationship is real, yet modest. It becomes more important later in a career, once people are trusted with independent decisions, ambiguous problems, and work that cannot be reduced to a checklist.

That makes sense.

Early jobs often compress differences. The new employee follows instructions, completes defined tasks, and earns roughly what others in the same role earn. Over time, some people are given broader problems: build the system, manage the uncertainty, make the judgment call, decide which information matters.

At that point, the ability to reason through unfamiliar complexity becomes more valuable.

But markets do not reward intelligence alone. They reward scarcity, timing, risk tolerance, social skill, access to capital, reputation, ownership, negotiation, and luck. A highly intelligent person can waste their ability. A less intellectually gifted person can build a strong life through discipline, good judgment, relationships, and persistence.

Intellect expands the range of doors a person can enter. It does not decide whether they walk through them.

AI and judgment

Artificial intelligence is changing this equation because it can now perform an early version of much cognitive work.

It can draft, summarize, translate, retrieve information, write code, generate plans, and turn a blank page into something that resembles a first answer. These are not trivial capabilities. They reduce the cost of production across much of knowledge work.

AI does not remove the need for judgment. It makes the absence of judgment scalable.

The central question is not whether AI makes people smarter or dumber. It is whether it helps them extend a skill they already possess, or allows them to avoid developing that skill in the first place.

Early research gives reason for caution. In one MIT Media Lab study, people using language models during essay writing showed lower neural engagement, weaker recall of what they had produced, and less ownership over the final work than people who used search or worked without external support. The study is preliminary, but its mechanism is familiar: when a tool performs the difficult step before a person has learned to perform it, the output arrives without the capacity to evaluate it.

The opposite can also be true. A notebook, calculator, map, spreadsheet, or language model can expand what a capable person is able to do. The difference is not whether the tool is external. The difference is whether the user remains responsible for the standard.

AI is leverage when it helps a person extend judgment.
It is substitution when it gives them a plausible answer they cannot assess.

That distinction may become one of the most important forms of inequality in the next decade. The meaningful divide will not simply be between people with more or less raw cognitive ability. It will be between people who use AI to raise their standards and people who use it to lower them.

The practical conclusion

The debate around intellect usually produces two failures.

The first is denial: the claim that cognitive differences are too uncomfortable to acknowledge, and therefore should not shape how schools, careers, or institutions are designed. This leaves people with generic advice that ignores what work actually demands of them.

The second is fatalism: the belief that a score is a life sentence. This mistakes a useful measurement for a complete person. It ignores uneven strengths, interest, discipline, opportunity, health, social context, and the fact that every individual score contains uncertainty.

Neither position is correct.

Cognitive ability is real. It is partly inherited. It is independent of character and only partly predictive of success. It matters more when work is novel, abstract, and constantly changing. Conscientiousness matters differently: it determines whether a person can reliably turn capacity into action.

A society that takes both dignity and function seriously should build around that reality.

It should offer more than one respected route into adulthood. It should value technical competence, practical skill, care work, craftsmanship, entrepreneurship, and skilled trades alongside academic and professional paths. It should help people find work whose complexity they can genuinely carry. And it should build AI tools that strengthen judgment rather than quietly replace the effort through which judgment is formed.

The task is not to argue human variation away. It is to design paths where variation becomes contribution rather than exclusion.