Now I live at the very peak of clarity and beauty, the existence of which I did not even suspect. Ideas explode like fireworks in my head. There is no greater pleasure in the world …
"I have always believed that von Neumann with his brain belongs to some other species, that this is a clear example of human evolution." (Nobel laureate Hans Bethe)
“Alpha kids wear gray. Alphas have a much harder job than ours because alphas are incredibly smart. It's wonderful that I'm beta, that our job is easier. And we are much better than gammas and deltas. The gammas are stupid. " (Brave New World by Aldous Huxley)
Nobel laureate and one of the founding fathers of the great school of Soviet physics, Lev Landau, had a logarithmic scale for ranking physics theorists with levels from one to five. A physicist of the first level was ten times more influential than a physicist of the second, and so on. He modestly set himself on level 2, 5, and only at the end of his life moved to the second level. On the first level, he had Heisenberg, Bohr and Dirac, as well as several other people. Landau put Einstein on the level 0, 5.
My friends in the humanities and other sciences such as biology are amazed and alarmed that physicists and mathematicians can think in such hierarchical categories. Obviously, in these scientific fields, the difference in ability is not so pronounced. But Landau's scheme seems to me quite appropriate: there are many physicists whose contribution I do not understand at all.
I even came to the conclusion that the Landau scale can, in principle, be extended beyond the 0.5 level, on which Einstein stands. Genetic studies of cognitive abilities indicate that today there are varieties of human DNA, which, if ideally combined, can lead to the emergence of individuals with an intellect that is qualitatively higher than anything that previously existed on earth. Roughly speaking, if we think in terms of the Landau scale, we are talking about people with an IQ of the order of 1000 points.
In Daniel Keyes's novel Flowers for Algernon, a mentally retarded protagonist named Charlie Gordon participates in a surgical experiment to improve intelligence, which raises his IQ from 60 to 200. From a bakery worker who was laughed at by friends, he turns into a genius. effortlessly understanding the many hidden connections in the world. “Now I live at the very peak of clarity and beauty, the existence of which I never knew existed,” writes Charlie. - Ideas explode in the head like fireworks. There is no greater pleasure in the world … This is truth, love and beauty, fused together. This is a delight. How can I give up all this? Life and work - better than this a person cannot have anything. The answers are already inside me, and soon, very soon they will burst into my brain. " The difference between superintelligence and today's average IQ of 100 will be even greater. The possibility of superintelligence arising is a direct result of the genetic basis of intelligence. Traits such as growth and cognition are governed by thousands of genes, each with a small impact. The approximate lower bound on the number of common genetic variants that affect each trait can be inferred from the positive or negative impact on it (height is measured in inches, and IQ in points) of already discovered varieties of genes, called alleles. The Social Science Genome Association Consortium, which includes dozens of university laboratories, has identified several sections of human DNA that affect cognition. They show that a number of snips (single nucleotide polymorphism, or DNA sequence differences of one nucleotide size) in human DNA are statistically correlated with intelligence, even after correcting for repeated tests of 1 million independent DNA regions in a sample of more than 100 thousand people. If cognitive abilities are controlled by only a small number of genes, then each of the gene varieties should significantly change the IQ - by about 15 points when comparing two people. But the biggest difference scientists have been able to identify to date is less than one IQ point. The big difference would have been easier to spot, but it hasn't been found. This means that there must be at least thousands of alleles for real differences to be observed in the general population. A more complex analysis (with a large margin of error) gives the final figure of 10,000. Each genetic variant slightly increases or decreases cognitive abilities. Because cognition is defined by the cumulative array of small spillover effects, it is usually dispersed and follows the familiar bell-shaped curve, with more people in the middle than at the edges. A person whose number of positive options (increasing IQ) is above average will exceed the average in ability. The number of positive alleles is higher than the average required to increase the value of a particular trait within the standard range, that is, by 15 points, in proportion to the square root of the number of variants, that is, equal to about 100. In short, a hundred additional positive varieties can increase IQ by 15 points. And since there are thousands of potential positive options, the conclusion is quite understandable. If a person can be genetically engineered to have a positive version of each causal variation, the result may be cognitive abilities that are about 100 standard deviations above the average. And this corresponds to more than 1000 IQ points.
It is completely incomprehensible what exactly the value of IQ will have within such limits. However, we can confidently say that whatever this value may be, this kind of ability will far exceed the maximum intelligence of any of the 100 billion people who have ever lived on earth. We can imagine the abilities of great scientists, who in their maximum form will be present all at once and in one person. This is an almost perfect reproduction of images and speech, ultra-fast thinking and calculations, powerful geometric visualization, and in higher dimensions, the ability to simultaneously and simultaneously perform many analytical and mental actions. The list goes on. Charlie Gordon, but squared.
To achieve this maximum, it will be necessary to directly adjust the human genome, creating favorable options for each of the 10,000 locations. In an optimistic scenario, this will someday become possible if gene-altering technologies like the recently discovered CRISPR / Cas system, which have sparked a revolution in genetic engineering in the past year or two, emerge. Harvard genomist George Church even suggested that CRISPRs (short palindromic repeats, regularly spaced in clusters) would allow mammoths to be revived by selectively modifying the genomes of the Asian elephant embryo. If Church is right, we should include supergenies in addition to mammoths in the list of wonders of the new genomic age.
Some of the assumptions behind the 1,000 IQ projection are now the subject of debate. To some, the very idea of quantifying intelligence seems controversial.
In his autobiographical book You Are Surely Joking, Mr. Feynman! Nobel laureate physicist Richard Feynman devoted an entire chapter to his attempts to avoid studying the humanities. He called her "Always Trying to Get Out." While studying at the Massachusetts Institute of Technology, he wrote, “I was only interested in science; nothing else worked for me."
Familiar moods. Conventional wisdom sometimes says that good mathematicians are at odds with literature, and vice versa. This distinction has influenced our understanding of genius, indicating that ability and giftedness are manifested in one part of the brain, but not as a whole. Because of this, the very idea of an IQ of 1000 points becomes problematic, since it is impossible to grasp the immensity.
But psychometric research, whose purpose is to determine the nature of intelligence, paints a completely different picture. Observations in the millions show that virtually all "primitive" cognitive abilities, such as short-term and long-term memory, the use of language, quantities and numbers, visual representation of spatial relationships, pattern recognition, and so on, are in a positive relationship and relationship.
Positive relationships between narrowly focused abilities indicate that a person with outstanding ability in one area (for example, in mathematics) is likely to have abilities above average in another (speech ability). They also show that there is a reliable and useful method of compressing information related to cognitive abilities.
There is another assumption about the 1000 IQ prediction, which is that cognition is highly influenced by genetics, and therefore it can be inherited. This assumption has very solid evidence. Behavioral geneticist and twin researcher Robert Plomin argues that the genetic influence on intelligence is stronger than on any other human characteristic.
In studies of twins and adopted children, pairing IQ ratios are roughly proportional to the degree of relationship, defined as the proportion of genes shared by two individuals. Only small differences were found due to the family environment. Children of the same parents who have no biological relationship, growing up in the same family, have almost zero correlation in their cognitive abilities. These results are consistently supported by other major studies conducted in different places, including in different countries.
It would seem that in the absence of hunger and deprivation, the upper limit of cognitive abilities is determined by the genetic effect. However, in other studies, where participants experienced additional environmental pressures, such as poverty, malnutrition, lack of education, heritability rates were much lower. Under unfavorable environmental conditions, a person does not fully reveal his potential.
Probably, the superintelligence is a matter of a distant future, but in the near future we can expect even smaller, but still important events. Large datasets of human genomes and their corresponding phenotypes (these are physical and mental characteristics of a person) will significantly expand our understanding of the genetic code, and in particular, our ability to predict human cognitive abilities. Detailed calculations indicate that millions of phenotypic-genotype pairs will be required to elucidate the genetic architecture using the most modern statistical algorithms. However, as the cost of genotyping is rapidly declining, this could happen in the next ten years or so. If the existing estimates of heritability suggest something, then the accuracy of predicting intelligence based on the genome can be better than half the standard deviation (that is, better than plus or minus 10 points).
When predictive models become available, they can be used in reproduction. This is the selection of embryos (selection of a fertilized egg for implantation), and active genetic modifications (for example, using CRISPR methods). In the first case, parents, choosing one of a dozen eggs, will be able to increase the IQ of their child by 15 or more points. And this is a big difference: either your child is barely doing well in school, or he goes to college and is successful there. Oocyte genotyping is already technically quite well mastered, and now all that remains is to develop a comprehensive phenotype prediction for embryo selection. The cost of this operation will be lower than the fees for many private kindergartens, and the consequences for a person will be for life, and for his descendants.
But there are corresponding moral issues that deserve close attention, and they will have to be solved in the rather short period of time that remains before such opportunities appear. Each society must decide for itself where to draw the line for genetic engineering in humans. And here we have very different prospects. Some countries are likely to allow this kind of genetic engineering, opening doors for the world's elite who can afford to travel abroad to reap the benefits of reproductive technology. As with most technology, the rich and powerful will be the first to benefit. But I believe that over time, many countries will not only legalize human genetic engineering, but also make it part of their national health care system (voluntary part).
The alternative would be inequality of a kind never seen before in human history.
Stephen Hsu is vice president of research at the University of Michigan and professor of theoretical physics. He is also a scientific advisor to BGI (Beijing Genomics Institute) and founder of its cognitive genomics laboratory.