Reply to Elon and Sriram

No, there are not millions of geniuses clamoring to come to the U.S.

Sriram Krishnan, Senior White House Policy Advisor on Artificial Intelligence (Wikicommons - TechCrunch).

If humans vary in their capacity for intelligence, why can’t they also vary in their capacity for empathy, guilt proneness, or willingness to follow rules? These mental traits are likewise heritable, and they likewise vary in adaptive value, being more valuable in some cultures than in others (Frost, 2020).

Perhaps you believe that the above traits form a single package. Perhaps you believe that intelligent people are normally good. Yes, cunning psychopaths do exist, but they’re the exception … aren’t they?

That belief is true, up to a point. As human groups increased in size and complexity, they underwent selection not only for intelligence but also for traits that help people trust each other, get along with each other, and interact peacefully.

However, some human groups have gone farther in that direction than others. Some have created truly large and complex societies, to the point that the sphere of social trust mostly encompasses people who are neither friends nor close kin.

Large high-trust societies have arisen only in Europe, especially northwestern Europe, and in East Asia. In both regions, we see similar trajectories of mental evolution: higher capacity for intelligence, less willingness to use violence to settle personal disputes, and greater willingness to follow and enforce rules. But there are differences. East Asians have created “shame cultures” — wrong behavior is discouraged largely by the feelings of shame you experience when other people know you have broken a rule. In contrast, Europeans have created “guilt cultures” — wrong behavior is discouraged largely by the feelings of guilt you experience after breaking a rule, even when you are the sole witness.

In a guilt culture, high levels of guilt-proneness co-exist with high levels of empathy, particularly affective empathy — you see someone in pain, and you begin to feel that pain. All human groups have some capacity for affective empathy, but most express it primarily in the mother-child relationship. The generalization of affective empathy to other relationships has been critical to the formation of high-trust societies in Europe.

Europeans thus control their behavior much more through internal, “hardwired” mechanisms: guilt to discourage wrong behavior and affective empathy to encourage right behavior. East Asians rely more on external, “softwired” mechanisms: not only shaming by other community members but also learning of empathy — one has to learn to be compassionate toward others, and this Confucian virtue (known as ren) is considered to be key to social harmony.

Of course, high intelligence has not evolved solely in the large high-trust societies of Europe and East Asia. It has evolved elsewhere, typically in small groups that trade with a much larger one while feeling no special responsibility for its wellbeing. Such groups have to adapt to the cognitive demands of trade — literacy, numeracy, planning, budgeting — but this adaptation doesn’t require turning the entire space of social interaction into a high-trust space. Trust has to be maintained only in smaller spaces, particularly the buyer-seller relationship. Although traders may perform acts of philanthropy for their host society, such acts are done to create a friendly climate and are not involuntary acts of guilt or empathy.

In sum, high intelligence does not act alone in creating a high-trust society. It acts in combination with other factors that channel it in the right direction, specifically by constraining behavior through high levels of empathy, guilt, and rule following. Without those constraints, high intelligence may do more harm than good.

Krishnan’s proposal for tech worker immigration

This brings us to a recent proposal by Trump’s senior AI advisor, Sriram Krishnan, to remove country caps on green card applicants. This would have the effect of greatly increasing tech worker immigration from India — the main source of foreign workers in Silicon Valley.

Krishnan’s proposal has been backed by Elon Musk in several tweets on X:

OF COURSE my companies and I would prefer to hire Americans and we DO, as that is MUCH easier than going through the incredibly painful and slow work visa process. HOWEVER, there is a dire shortage of extremely talented and motivated engineers in America.

In a later tweet, Musk added: “Maybe this is a helpful clarification: I am referring to bringing in via legal immigration the top ~0.1% of engineering talent as being essential for America to keep winning.”

Is Musk being realistic? Can the U.S. recruit large numbers of foreigners who are as smart as the top 0.1% of American engineers? Let’s examine his goal point by point:

• Engineers have a mean IQ of about 110 (Wolfram, 2023).

• The top 0.1% of Americans have IQs of 147 or more. Therefore, the top 0.1% of engineers is ten points to the right, i.e., IQs of 157 or more.

• Very few people are that intelligent. Global mean IQ is estimated at below 90 (Gouillou, 2024, pp. 147-152). Even if we assume a global mean of 90, we still get a very small number of people with IQs of 157 or more, perhaps a little over half a million (Gouillou, 2025). Most of them already live in affluent countries.

It is delusional to think that millions of geniuses are clamoring to come to the U.S. Foreign tech workers will overwhelmingly be average people with IQs of around 110. That level of intelligence is consistent not only with the mean IQ of engineers but also with a proposal that Trump made last June:

What I want to do, and what I will do, is you graduate from a college, I think you should get automatically, as part of your diploma, a green card to be able to stay in this country. And that includes junior colleges too. Anybody graduates from a college — you go in there for two years or four years. If you graduate, or you get a doctorate degree from a college, you should be able to stay in this country (Sullivan, 2024).

Trump’s proposal was likewise greeted with happy talk about bringing “the best and the brightest” to America. In reality, a college degree no longer attests to high intelligence. American college students are now just above the cognitive average, with an estimated mean IQ of 102 (Uttl et al., 2024).

Parting words

It would be simpler to administer an IQ test to prospective immigrants and reject any who have a score less than 110. Unfortunately, American jurisprudence forbids using IQ tests if the results lead to discrimination on the basis of race, color, religion, sex, or national origin (Wikipedia, 2024). In this case, the results would be highly discriminatory: there would be almost no immigration at all from large parts of the world. This is why many immigration reformers have pushed for educational requirements as a proxy for IQ — in the mistaken belief that college graduates have above-average intelligence.

Even if one could find a legally acceptable way to select prospective immigrants for intelligence, the results would still be disappointing. It is not simply that the world has fewer geniuses than some of us think. There is also the need to screen for other mental traits that are no less important for the survival of a high-trust society. America’s success has not been due primarily to its citizens being smarter than the global average. The main reason is the country’s high level of social trust — a result of equally high levels of empathy, guilt proneness, and rule following.

Unfortunately, those traits are even scarcer than high intelligence in today’s world. High levels of empathy and guilt-proneness have been favored only in guilt cultures, and those cultures are a shrinking fraction of the world’s population. High levels of rule following are less scarce, but still far from universal. Also, most human groups apply their rules in nepotistic ways: while wrong behavior toward a close relative, especially a family member, is severely punished, wrong behavior toward a non-relative is less serious and often tolerated. This may seem unfair, but it’s how most of the world works.

Americans love to play fair. To them, it seems only “fair” that immigration should be based on a single metric that applies equally to everyone — in this case, a college degree. Meanwhile, it also seems “fair” that everyone should get a college degree, with the eventual result that a college degree no longer means much. In this, and in many other ways, “fairness” is piling up one social contradiction on top of another. At some point, Americans will have to accept the limitations of fairness and the reality of human differences.

At present, the best solution would be to let in as few immigrants as possible. Any admission criterion, no matter how “fair” — criminal record, university transcript, letters of reference — will prove to be worthless, either because it can be faked or because it simply fails to measure the innate characteristics that prospective immigrants will pass on to their descendants.

References

ABC News (2024). Musk, Ramaswamy spar with Trump supporters over H-1B work visas. December 28. https://www.msn.com/en-us/politics/government/musk-ramaswamy-spar-with-trump-supporters-over-h-1b-work-visas/ar-AA1wDydt?ocid=BingNewsSerp

Frost, P. (2020). The large society problem in Northwest Europe and East Asia. Advances in Anthropology, 10(3), 214-134. https://doi.org/10.4236/aa.2020.103012

Gouillou, Philippe (2024). IQ: From Causes to Consequences. From genetics to cognitive capitalism. France: Gouillou.com. https://www.amazon.ca/IQ-Consequences-Genetics-Cognitive-Capitalism/dp/295939853X

Gouillou, Philippe (2025). Worldwide Distribution of IQs. https://osf.io/preprints/psyarxiv/6g7bn

IQ Percentile and Rarity Chart. https://www.iqcomparisonsite.com/IQtable.aspx

Sullivan, K. (2024). Trump says he wants foreign nationals who graduate from US colleges to ‘automatically’ receive green cards, CNN, June 21. https://www.cnn.com/2024/06/20/politics/trump-green-cards-gradutate-college/index.html

Uttl, B., Violo, V., & Gibson, L. (2024). Meta-analysis: On average, undergraduate students' intelligence is merely average. ScienceOpen Research, May 1. https://doi.org/10.14293/S2199-1006.1.SOR.2024.0002.v1

Wikipedia. (2024). Griggs v. Duke Power Co. https://en.wikipedia.org/wiki/Griggs_v._Duke_Power_Co.

Wolfram, T. (2023). (Not just) Intelligence stratifies the occupational hierarchy: Ranking 360 professions by IQ and non-cognitive traits. Intelligence, 98, 101755. https://doi.org/10.1016/j.intell.2023.101755