In 1990 researchers reported the first successful use of preimplantation genetic testing. The embryos belonged to couples in which the female partner was a carrier of one of two heritable conditions that typically affect only males — one a form of intellectual disability, the other a rare deadly condition called adrenoleukodystrophy. The embryologist Alan Handyside, in a major breakthrough, took single cells biopsied from embryos and identified their sex, which allowed the couples to choose female embryos for implantation and avoid passing on those conditions.

Over the next three decades, advances across various fronts have led to more sophisticated and targeted testing. Today some form of preimplantation genetic testing, or P.G.T., is used in over half of I.V.F. cycles in the United States, at a cost of $3,000 to $5,000 per batch of embryos. The most common options patients have are tests for extra or missing chromosomes, structural chromosomal rearrangements that can trigger pregnancy loss and disorders linked to a single gene, such as cystic fibrosis and muscular dystrophy.

More recently, with the advent of powerful statistical techniques that can analyze huge databases of genetic information, several American companies have started offering P.G.T.-P., which screens embryos for their polygenic risk scores. The technology has typically been used for adults, ostensibly to assess their probability of developing specific conditions. For example, people whose test shows a high risk score for heart disease might change their diet or increase their physical activity.

The risk scores come from what are called genomewide association studies, which identify which of hundreds of thousands of genetic variants are statistically linked with a specific condition or trait. While tests for single-gene diseases can conclusively verify the presence or absence of a specific genetic mutation, these scores can assign only probabilities and do not account for environmental or other factors.

The usefulness of polygenic risk scoring in adults is still an open question; its application to embryos is even less straightforward. Because the results are probabilistic, having a slightly elevated risk of a condition does not necessarily translate into developing it. The risk calculus is further complicated by the vastly different environment for a child born today compared with the adults whose biological samples, the large-scale collection of which began in the mid-2000s, make up the data sets from which these risk scores are generated.

Todd Lencz, who researches genetic biomarkers for schizophrenia, gave me the example of a genetic variant that predisposes people carrying it to smoke more. Currently, polygenic risk scores take into account those genes’ presence as a predictor of shorter life span or higher risk for cardiovascular disease. But people born with that variant today will encounter an environment and a culture around smoking very different from those of someone who grew up in the 1970s with the same predisposition. Smoking-prone babies born now might “never even encounter a cigarette in their life,” he said.

There are many other questions about P.G.T.-P.’s accuracy and efficacy. Much of the analysis that has generated the risk scores comes from two large data sets, one American and one British. Because of the demographics that make up these data sets, the screening tools are most accurate for people of European descent. Other countries, particularly in East Asia, have assembled their own biobanks, and a multicountry project across Africa has resulted in the discovery of millions of variants in the human genome. But how these findings apply to people with parents of, say, multiple ethnic backgrounds remains “a wide open question right now,” Dr. Lencz said.

Advocates of the screening argue that it nevertheless provides valuable information that can help parents assess their embryos’ propensity to develop certain diseases: “Have healthy babies,” as the landing page of Orchid puts it. But polygenic embryo screening goes further than the dubious promise of health. Studies have identified sets of genes linked to everything from educational attainment and height to mental health conditions such as depression and schizophrenia. It’s one thing to screen for conditions like Type 1 diabetes; it’s quite another to go looking for the embryo deemed most likely to clear six feet and test into the Ivy League.

In July the technology newsletter The Information reported on Silicon Valley’s enthusiastic investment in and consumption of frontier fertility technology. It cited three sources who said that in private, Ms. Siddiqui has “claimed that her company is able to measure embryos’ intelligence.” (A company representative denied the claim to The Information.)

Genomic Prediction, a New Jersey-based testing lab that offers a range of screening services, including polygenic screening, initially offered consumers a score for gene variants associated with intellectual disability but rescinded it after it attracted controversy. (A company representative said it was removed from the panel of markers it can test for after the screening was found to “not be effective, given the data.”) “The testing was meant to look at intellectual disability, not intellectual ability,” Nathan Treff, Genomic Prediction’s chief scientific officer and one of its founders, told me, disputing the idea that parents were looking for supersmart embryos. Intellectual disability, he stressed, is “a medical condition,” and he did not rule out one day offering screening for it again.

This is technically true — but the line between screening for a disability and screening for IQ is not as clear as he makes it out to be. Someone with an IQ of around 70 is considered to meet the threshold for an intellectual disability. But for the cases in which that intellectual disability is not caused by a known genetic syndrome, the technology is not accurate enough to screen out an embryo that will grow up to have an IQ of 69 — intellectually disabled — versus one with an IQ of, say, 78.

Thus, the only way to screen for intellectual disability is through screening for IQ more broadly by combing through the many hundreds or thousands of DNA markers that contribute to intellectual ability — a less P.R.-friendly truth that some in the sector downplay to the public at large — even as others openly market it to potential clients. In October, The Guardian, which had reviewed undercover recordings obtained by the British-based nonprofit Hope Not Hate, reported that a U.S. company, Heliospect Genomics, claimed in a presentation to prospective clients that it could select the “smartest” out of 10 embryos, for an average gain of more than six IQ points.

The same dynamics apply to other conditions. While these companies say they’re screening for idiopathic short stature or psychiatric conditions like depression, it is not at all clear that they have the ability to deliver on the precision promised. Does a high risk score for idiopathic short stature mean an embryo will grow up to meet the threshold for a medical condition? Or simply that the person will be on the shorter side? Does a high score for depression mean that a given embryo might grow into a person with mild seasonal affective disorder or develop debilitating mental illness?

Some people who study genomics don’t believe the technology, now or in the future, can deliver on its promises, in part because many diseases are caused by numerous risk factors beyond genetics. “The risk from things the test can’t measure is much greater,” said Anneke Lucassen, a clinician and a professor of genomic medicine at the University of Oxford. Using such incomplete information to make decisions about which embryo to implant is “about as likely as a coin toss to deliver the outcome desired.”

Polygenic embryo screening has not yet gained widespread acceptance, but one representative survey of Americans conducted in 2022 found that a majority said they had no moral objection to using P.G.T.-P. for medical and nonmedical traits and nearly 4 in 10 said they were “more likely than not” to use it if it would slightly increase their child’s likelihood of getting into a top college.

Another recent study found that while potential patients were enthusiastic about the possibilities, medical providers were not persuaded. “We’re not here to create what we think is or what the patient thinks is ideal. We’re here to help them with a medical condition, which is infertility,” one fertility doctor who was surveyed said. Other care providers surveyed raised concerns about lawsuits. What if patients selected embryos that were supposed to be “tall, beautiful and smart and they’re short, squat, thick and a little dull?” one wondered.

In December 2023 the Psychiatric Genomics Consortium, an international group of researchers studying the biology of mental health, criticized Orchid for using its data; the consortium’s founder made the point that the group’s aim was to improve the lives of people with mental illness, not to prevent them from being born. (An Orchid spokesperson said the company did not use the consortium’s data but did not share its source.)

New technologies affect not only patient preferences but also social expectations around embryos and how we treat them. Vardit Ravitsky, a bioethicist and the president of the Hastings Center, which examines ethical issues related to health and technology, sees these new screening tools as a step change from previous iterations of genetic testing. “It’s a qualitatively new message of: We should have the best children that we can, across the spectrum,” she said. Once such a tool is available, “it immediately becomes a societal expectation to use it, and the rejection of it or the refusal of it becomes a morally significant choice,” she continued, pointing to the extensive literature showing that women who refuse prenatal testing are seen as irresponsible.

I knew I was watching a sped-up video, but I consumed it at face value nonetheless. I watched the two pronuclei appear, then fade away. At 30 seconds in, the circle split into two blobs that jostled for space. Then there were four cells, then eight. The cells got smaller each time they divided. By a minute or so in, the mass of cells began to resemble a rosette of wobbly orbs — and a lively one indeed.

On the one hand, it was hard to look at the image and not acknowledge that what I was looking at was, to use a timeworn expression, a clump of cells. They were cells. And they were arranged in a clump. At the same time, I knew the fate of this specific clump of cells: They had been observed over five days by our embryologist and stood out to her as our best chance for a child. They were selected, transferred and then, miraculously, implanted. They continued to develop and specialize over the course of nine months into organs, a spine, fingers, toes. Finally, they — all trillion-plus cells of her — emerged into the world.

At two minutes into the video, the clump, my daughter, seemed to enlarge slightly, and the hollowing out, lumenogenesis, began. A few little cells slipped out of the circle and danced around on the surface of the zona pellucida, hovering near the spot where they’d escaped, like an astronaut just outside the spacecraft. My mind was awash in clichés. I was in awe, transfixed, mesmerized.

Watching her develop onscreen was a novelty for me, because I already knew the outcome. But not all of it: My daughter isn’t an Orchid baby. Her genetic risk factors remain a black box. I don’t yet know what, if any, health conditions she will face or whether I could have done something about it when she was an embryo. Now that she and I are acquainted, it seems insane — to borrow Ms. Siddiqui’s term — to reduce her to a string of genetic code, given the delightful person that she is. But there will always be a part of me that wonders, guiltily, if I did her a disservice by not learning more.

Some argue that these technologies are simply options, ones we are free to use or not. But regardless of whether we use them, their mere existence will alter our relationship to embryos. By animating embryos and enabling a close-up encounter with them, time-lapse microscopy renders them more human than they already are; polygenic embryo screening reduces them to a set of probabilities. I am not convinced that not using such technologies leaves one untouched.

Right now, the availability of polygenic embryo screening is dictated by the market, its promise most vocally endorsed by the tech elite, who, as backers of the companies selling it, may have a vested interest in promoting it. The ability to act on the information it purports to provide — to hire a personal trainer, switch to an all-organic diet or consult health specialists who don’t take insurance — is also reserved for those with means. One doesn’t have to fully buy into its promises to worry about its implications, to wonder if this is a space in which market forces should perhaps not be given free rein and to feel that, whether or not we avail ourselves of polygenic embryo screening, it and related innovations will eventually have implications for us all.

Look for the third and final chapter of this series when it publishes next week.

Anna Louie Sussman is a journalist who writes about gender, economics and reproduction. She is working on a book about family building in an age of uncertainty. This project was supported by the Pulitzer Center and the Alicia Patterson Foundation.

Additional reporting by Tenzin D. Tsagong.