Optimized Offspring

Brigida D’Agostino is a Los Angeles-based news anchor and journalist covering what happens when technology changes the arrangements that shape everyday life, from reproduction and work to family and government. Five years ago, I sat in my OB-GYN’s office, three months pregnant, reading a stack of consent forms for noninvasive prenatal testing (NIPT) to find out whether my unborn daughter might have chromosomal abnormalities that would come with lifelong health problems. I wanted the test because I was 36 and my pregnancy was considered “geriatric” — the health system trying to convince me that something was wrong. I wiped the ultrasound goo off my still-flat belly and scribbled my signature on the forms. By the time my daughter was born, a newer kind of genetic screening service had reached the market: polygenic embryo screening, which ranks embryos developed using in vitro fertilization (IVF) by adding up thousands of tiny genetic variants to estimate the risk of disease or the likelihood of traits in ways that were not available before. The first company to commercialize the service was Genomic Prediction through its LifeView test in 2019. Sam Altman was an early investor. Nucleus Genomics is another company doing similar work. These companies’ marketing suggests that, for a price, you can design and optimize your child before embryonic implantation. The company Orchid Health charges $2,500 to screen each embryo. That’s on top of one IVF cycle, which starts at around $15,000. Nucleus bundles embryo screening and IVF into a plan with a four-month minimum, costing about $40,000. The Nucleus website advertises its services as “Modern IVF to give your baby the best start.” The company’s tests screen parents for thousands of genetic disorders and traits they could pass to their children, and scan embryos for diseases and other characteristics. An image of its web app shows a report card on a mobile phone titled “Kids’ most likely traits.” The traits are listed like settings. Height: 5 feet 8 inches. Eye color: hazel. IQ: +2 points. BMI: below average. Muscle strength: average. The American Society for Reproductive Medicine’s ethics and practice committees concluded in December 2025 that polygenic embryo screening is unproven and should not be used clinically at this time. Nucleus’s FAQ says it is “built on validated, peer-reviewed research” and already used in IVF clinics across the U.S. and globally. Today, doctors and scientists can screen embryos for more traits than ever before. Researchers are also getting better at editing their genes. In June, Columbia University researchers reported in a preprint (not yet peer-reviewed) that they had edited early human embryos with what The New York Times called “unprecedented accuracy” and far less chromosome damage than earlier CRISPR-Cas9 experiments. The technique, called base editing, isn’t ready for reproductive use yet, according to Dieter Egli, the Columbia geneticist who led the research. Nucleus’s chief clinical officer co-authored the study. Silicon Valley’s interest in reproductive technology is only growing. Altman has backed Preventive, a startup researching embryo editing. Peter Thiel’s Founders Fund and Anne Wojcicki, a co-founder of 23andMe, have also invested in embryo screening firms. Imagine: report cards for two embryos, each hinting at the person they might become. One apparently has an elevated risk of the melanoma that runs in your family. The other is more likely to share your green eyes. A mini you. Which do you choose? I asked myself this question and made my choice in seconds, but later it started to bother me, even though it was a hypothetical. What Culture Does For Evolution Humans adapt through two systems. We adapt through biological evolution, which allows us to inherit genes slowly over generations. And we adapt through cultural evolution, which lets us inherit information quickly so we can feed ourselves, build houses and learn language without waiting for new genes to arrive. Arizona State University anthropologist Charles Perreault’s 2026 study found that cultural evolution allowed humans to colonize the globe in less than 1% of the time it would have taken through biological evolution alone. Culture, he argued, “enabled the rapid and cumulative acquisition of adaptive behaviors and technologies” that allowed humans to thrive “while remaining a single species with low genetic diversity.” I wondered whether these newer reproductive technologies might let cultural preferences change human evolution on a species-wide scale. I also wondered whether they might create a new genetic class, where some families can buy reproductive advantages that other families can’t. Joseph Henrich, a prominent theorist of gene-culture coevolution, wasn’t convinced. He told me humans have always used cultural preferences to shape the gene pool, just slowly and indirectly over generations. “When we pick a mate, we pick the genes we like,” he said. Humans are incredibly good at reading traits through observation. We watch how a person we’re romantically interested in behaves, how they look. “We are much better than a polygenic score at predicting what parents care about.” But the market is built to speed up that dynamic and hand it off first to the people who can pay. IVF has already added another point of selection by allowing parents to choose among embryos, but for years there was little to choose because the data was limited. Polygenic screening gives parents much more information than was previously available, including predictions related to appearance, intelligence and health — which are increasingly being framed around optimization rather than treatment. “Embryo selection is super crude because the genes in any polygenic scores are also in scores for known traits and tons of other unknown traits,” Henrich said. Data for these predictions is also drawn largely from people of European ancestry. Guidance published in The American Journal of Human Genetics in 2022 says a score’s accuracy drops when applied to people outside the dataset it was built on. Henrich also said selection won’t have lasting genetic consequences unless the wealthy somehow sealed off their mating pool and married only within it. But the market doesn’t require endogamy to create a genetic class. It only needs enough people who can pay for choices that others leave to chance. Some investors and industry figures are already describing embryo selection in class terms. A partner at Peter Thiel’s Founders Fund who invested in Nucleus told The Washington Post that embryo selection is a “false choice” today but “in the future, it’s the trust fund.” The Wall Street Journal reported that embryo screening company Herasight’s in-house philosopher said repeated selection over generations could produce an intelligence gap between the “genetically enhanced” and “genetically unenhanced.” “There’s going to be inequality,” he said. Perreault told me embryo selection is still gene-culture coevolution, but it creates “an unusually tight feedback loop between cultural preferences, reproductive decisions and genetic transmission.” “Life needed 4.54 billion years to find a way to edit itself,” he continued. Once reproductive technologies begin influencing which embryos are implanted, he added, “some cultural choices become biologically ratcheted into future generations.” He compared it to fertility decline and the spread of contraception, neither of which was designed as a population-level evolutionary intervention. But both changed reproductive patterns over generations. Embryo selection, he said, would be “many individually rational decisions that, in aggregate, could reshape the evolutionary composition of a population in ways that no single actor intends or controls.” He said the magnitude of this effect would depend on how widely embryo selection is adopted and which traits parents select for. Screening lets parents select among embryos without altering them, picking the best of what chance produced. But editing would let parents modify an embryo’s DNA and introduce sequences neither parent carried. What Else Are We Choosing? The problem is that we don’t really know what we’re choosing. Genes are more complex than a report card suggests. A single gene or genetic variant can shape several unrelated traits at the same time. Geneticists call it pleiotropy. A trait that looks like a benefit in one place can carry a liability somewhere else. A 2018 study of more than 3,700 Mensa members found higher self-reported rates of anxiety, autism, ADHD, allergies, asthma and autoimmune conditions than in the general population. The researchers called it the hyper-brain, hyper-body theory. It doesn’t prove high IQ causes those conditions, but it shows how they may be linked. I’m a good example of why pleiotropy matters. I’m “twice-exceptional,” which means I’m both autistic and intellectually gifted. My life is harder because of this. I get hives, I’ve fainted from anxiety and I have the gift and curse of monotropism, an extreme form of focus on one interest at the expense of everything else. I’ll ignore my needs for food, water, sleep and the bathroom until my body gives out: my stomach growling, hands shaking and eyes burning. I’ll finish a month’s worth of work in two weeks, but my husband and daughter won’t be happy about it. These traits may be linked in ways scientists don’t fully understand. What would my genetic report card have told my parents about me? What traits would have been detected and how would they have been ranked? Would they have seen a mind of high potential ability or someone at risk of anxiety? Given a choice, would they even pick me? Even if they did, they might not have gotten what they paid for. We may think we’re in control, but, Henrich told me, “It’s far from clear that all that ‘rational’ selection will do anything other than add more randomness.” Editing genes we don’t fully understand also carries a risk. In 2018, news broke that a Chinese scientist named He Jiankui created the first gene-edited babies. He used CRISPR to modify a gene called CCR5 in twin embryos, hoping to make the girls resistant to HIV, since their father was HIV-positive. But the gene may help fight other infections, including severe West Nile virus. The embryos didn’t have the HIV-resistant mutation that He Jiankui intended to introduce. But the changes he made can be passed on to any children the twins have. He Jiankui went to prison. What he did to those girls can’t be undone. Gene editing suggests that people can optimize their offspring, but we don’t know whether the editing will result in a trait or characteristic the parents or the person or society at large will value. There is no consensus on the optimal body and mind. The traits on Nucleus’s report card reflect the preferences of this moment. “The desirable traits of the 20th century, like IQ, will unlikely be those of most interest in the 21st century,” Henrich told me. The Moral Choice? A parent selecting an embryo in 2026 is making a choice for someone who may still be alive in 2086. Cultural preferences can change quickly, but reproductive decisions can have consequences that last generations. In 1980, the Chinese government implemented a policy limiting most families to one child. The government never told families whether to choose a boy or girl, but Chinese parents consistently selected for sons. One household at a time, it snowballed into a massive gender imbalance. Today, there are over 30 million more men than women. A large portion of surplus men might never have children because of the imbalance. In Denmark, prenatal screening resulted in a similar population change. In 2004, the Danish government became the first to offer free Down syndrome screening to all pregnant women. The government didn’t tell parents to end their pregnancies. But after screening showed a higher risk and follow-up testing confirmed the condition, most parents did. The number of children born with Down syndrome halved and has held steady ever since. Parents in both countries used prenatal tools that accurately showed them what they were looking for: a pregnancy with a boy or a girl, or a child likely to be born with Down syndrome. Embryo screening doesn’t offer that same certainty. Results are based on probabilities for traits that no one can guarantee, like a lower risk of depression. The other difference is price. Denmark’s screening was free, and China’s policy was a national directive. Embryo screening with IVF is expensive. We tend to draw moral lines around these choices as if our judgment changes the outcome. Many people think it’s reasonable to screen out Down syndrome or cystic fibrosis. Selecting for sex, intelligence, personality or appearance feels different. The industry draws these lines too. Genomic Prediction’s co-founder, Stephen Hsu, called some traits “too controversial.” And some companies, including Orchid, only screen for disease risk and medical conditions. But Denmark and China both reshaped their populations without meaning to, whether individuals made choices that society considered “moral” or not. How To Govern Genetic Engineering Orchid draws one line and Nucleus another, which means the commercial limit is set by the company that’s willing to go the furthest. Today’s laws were written before screening companies advertised on subway platforms, and before startups began shopping for countries where the rules around embryo editing are ambiguous. Preventive has explored the United Arab Emirates as a location for embryo editing, according to The Wall Street Journal. The company says its goal is preventing serious hereditary disease, not enhancement, and that it won’t pursue clinical use until safety is established. So far, the scientific and professional response has mostly been limited to warnings. In 2025, a coalition of gene-therapy societies called for a global moratorium on heritable editing. The consortium warned that selling germline modification for enhancement or parental preference echoes “the discredited ideologies of eugenics.” In the U.S., new reproductive tools and techniques often outpace current laws governing embryo research. Federal oversight is fragmented, and state rules vary. An annual appropriations rider prevents the FDA from accepting an application for a clinical trial involving an embryo intentionally modified to include a heritable genetic change. The bioethicist Henry Greely predicted a decade ago that people would start having their children in the lab, choosing among embryos even when they had no trouble conceiving. He told me the most obvious place to regulate is across about 500 IVF clinics where embryos are screened and implanted. He said regulating them “wouldn’t be easy,” especially since the FDA has almost no experience with them. Regulating genetics labs would likely be even harder and would raise serious questions about FDA jurisdiction. If the FDA lacks it, he said, no one else has it under current U.S. structures. In the U.K., the Human Fertilisation and Embryology Authority can approve or ban specific procedures. “It seems to do a pretty good job,” Greely said, though he noted it wields far more power than anything the U.S. has or is politically likely to get. Even so, a national regulator wouldn’t stop an intended parent or rogue scientist from going overseas. “I think international regulation with teeth is almost impossible,” Greely said. He Jiankui went to prison under a Chinese law that didn’t directly address editing because no such law existed. He was sentenced to just three years in prison. Legal ambiguity weakens punishment after an experiment has already been conducted and doesn’t stop future attempts. The UAE’s 2023 genome law restricts the modification of human traits in embryos while allowing genetic research aimed at diseases. The law is new, and its boundaries are untested. But Preventive has been exploring the jurisdiction, where a path toward eventual clinical use may exist that the FDA rider bars in the U.S. Greely estimated it would take a couple million dollars to set up an adequate facility for a few germline edits. Cheap enough, he said, that it could happen in countries without regulation. Greely said he sees few tools beyond “soft power condemnation and ostracism” where regulation doesn’t reach. He Jiankui became a global pariah due to a shared sense among scientists that editing a human before birth is unethical if we aren’t sure it’s safe. Much of the taboo rested on the technical dangers of CRISPR. The Columbia team’s base-editing results begin to weaken one of those safety arguments. There’s no consensus on when editing embryos for reproduction becomes safe enough to try. What We Think We Know When I started reporting this story, I worried that embryo screening created a simple way for culture to steer species-level evolution and tighten the feedback loop between culture and biological evolution. After talking with geneticists, anthropologists and bioethicists, I couldn’t prove it would change Homo sapiens. What I found was a tighter feedback loop than existed before, and a market sorting who can buy a potential genetic advantage. For 300,000 years, humans adapted to situations they couldn’t predict by carrying variation no generation was aware it would need. Most cultural preferences came and went faster than they could be written into DNA. Reproductive technologies narrow that gap. “It is no longer safe to assume, as some models do, that culture is merely external to biology,” Perreault told me. “The two systems remain distinct, but they interact more directly than they once did.” More than a decade ago, biologist Craig Venter, a key player in the effort to sequence the human genome, imagined a future in which technology could “speed up biological evolution to the pace of social evolution.” Heritable gene editing could push that acceleration further by turning a cultural preference into a genetic change in a single generation. Once a person is born from an edited embryo, that genetic change is part of their DNA and can be passed on. Reversing it in future generations would require another intervention plus the technology, money and motivation to do it. Embryo screening is more limited, but it can still influence which genetic traits a child inherits and may eventually pass down. Today’s reproductive genetics market is global with no effective international regulator, driven by decentralized decisions as new funding, scientists and companies move toward heritable edits. If embryo screening becomes accurate and effective in another decade, it may become as routine in IVF as NIPT in pregnancy. The slow variation of biology plus the fast turnover of culture would be traded for the preferences of a single generation. There has been no collective global decision about whether this is OK. Some families will make a choice and the rest will live with whatever consequences come. By using these technologies, we assume we can identify which traits will matter — whether the children we choose or their children’s children want that or not. A trait that looks like a benefit today could turn out to be a liability later. Today, we have an illusion of control over our genetics — we don’t know if our offspring will be improved by edits to their genes. I chose immediately and I have no idea if my choice was right. Even if the technology improves, I’m not sure the rest of us will be better off.