This type of genetic testing is possible thanks to multiple advances in technology—not just in genomics, but also in the ability to keep embryos alive in a dish for five to six days and the technique of freezing embryos while the cells undergo testing and thawing them once the results are in. And it has become hugely popular—some clinics do PGT-A tests on all their embryos.

But PGT-A won’t give you a perfect readout of a future baby’s genetics, says Sonia Gayete-Lafuente, a reproductive endocrinologist at the Center for Human Reproduction in New York City. And some of the abnormalities might be able to self-correct with time. Gayete-Lafuente and her colleagues have transferred some of those “abnormal” embryos into patients’ uteruses and seen them develop into perfectly healthy children, she says.

Other forms of PGT are even more controversial. PGT-P tests are designed to predict an embryo’s chances of developing complex traits that rely on multiple genes, including medical disorders but also physical characteristics like height or cognitive factors like IQ. These tests are new, and they are illegal in some countries, including the UK. But they are gaining ground in the US. Nucleus Genomics—a company that invites customers to “have [their] best baby”—promises to predict traits running the gamut from eye color and intelligence to left-handedness and risk of Alzheimer’s.

When I asked IVF practitioners how they might respond if a patient asked for this service, most dodged the question and told me there’s not enough evidence that any of these tests actually work. They also cautioned that selecting for one trait might inadvertently introduce new risks. None seemed especially keen on the idea of using genetic testing for anything other than preventing serious disease.

3. Speeding things up with AI

Some seemed more excited about the potential for AI. After all, AI tools are generally good at recognizing patterns. Many researchers have attempted to train tools to spot healthy sperm, eggs, and embryos.

And they’ve had some success. A team at Columbia University Medical Center in New York has developed a device that uses AI to examine semen samples from men who have only tiny numbers of healthy sperm. An embryologist might struggle to find a single healthy sperm in such a sample. But the Sperm Tracking and Recovery (STAR) system can analyze over a million microscope images in an hour. It has already been used to create healthy embryos. The team behind the work announced the first pregnancy resulting from the treatment in November last year.

Other teams are using AI tools to advance IVF in more dramatic ways. Around a decade ago, a reproductive endocrinologist named Alejandro Chavez-Badiola began developing an AI tool trained to rank embryos, another to rank eggs, and another to select sperm. He recalls being struck by a realization that these tools were “the brains that have the potential to drive robots in the future,” he says.

4. Using robots to standardize IVF

In the early 2020s, Chavez-Badiola and his colleagues decided to combine technologies and develop an automated system for IVF. In theory, a robotic system loaded up with AI tools could undertake most of the steps required in the IVF process: selecting the eggs and sperm, fertilizing eggs to create embryos, culturing those embryos in a dish, and selecting the “best” one for transfer. Such a system could “do everything in a standard way” without ever getting tired, he says.