AI-Assisted Genetic Diagnosis and Its Ethical, Legal and Social Implications: A Scoping Review
This scoping review maps the current landscape of ethical, legal, and social implications (ELSI) in AI-assisted genetic diagnosis, revealing that most existing scholarship fails to address the unique intersection of AI and genomics, thereby highlighting an urgent need for genetically-specific governance frameworks and fairness audits.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Imagine a future where a computer program can look at a person's genetic code and predict their risk of developing certain diseases, or help doctors understand why a patient has a rare condition. This is the promise of artificial intelligence in genetic diagnosis. However, human DNA is not just a list of health facts about one person. It is a shared family heirloom. A single person's genetic code reveals information about their parents, their siblings, and even their children who were not born yet. It also connects them to a larger group of people who share their ancestry. Because of this, the rules that apply to regular medical data do not fully fit genetic data. When a computer makes a mistake or a secret is leaked, the consequences ripple out to relatives and entire communities, not just the individual. As these powerful tools become faster and cheaper, scientists and ethicists are racing to understand the specific dangers and responsibilities that come with using them on our most personal biological information.
A new study by researchers at Harbin Medical University and Harbin Vocational College of Science and Technology takes a close look at how scholars are currently thinking about these issues. The team conducted a broad search of scientific literature to see what has been written about the ethical, legal, and social problems of using artificial intelligence for genetic diagnosis. They were not looking for a simple list of pros and cons, but rather a map of the entire field to find out where the thinking is strong and where it is missing pieces. The researchers searched through thousands of records from major scientific databases, focusing on the specific intersection where computer algorithms meet human genetics and ethical concerns. After removing duplicate entries, they screened 981 unique records. They found that 186 of these papers were relevant enough to study in detail, while the vast majority—795 records—were excluded because they did not address all three areas at once.
The researchers discovered that most existing discussions fall into one of two categories. Some papers treat genetic diagnosis as just another type of medical technology, applying general rules about computer ethics that ignore the unique nature of DNA. Others treat it as a problem of genetics alone, applying old rules about genetic privacy that ignore how modern computer learning works. The study found that the crucial middle ground, where the specific dangers of AI and genetics overlap, is largely empty. In fact, the data showed that 81 percent of the records the researchers found failed to address this intersection properly. They were looking for a conversation that combined the algorithm, the genome, and the ethical implications, but they mostly found two separate conversations happening in parallel.
To make sense of the 186 papers that did address the topic, the researchers organized the findings into six main themes. The first theme is about confusion and clarity. Because genetic data is so complex, it is hard to explain to a patient why a computer made a specific prediction. The study suggests that current methods for explaining these decisions often create a false sense of understanding, making patients feel they know more than they actually do. The second theme deals with responsibility. If a computer makes a mistake that harms a patient's family, it is unclear who is to blame. Current rules do not clearly assign responsibility for harm that spreads to relatives or future generations. The third theme focuses on privacy. Most privacy laws protect the individual, but genetic data can reveal secrets about a whole group of people. The study notes that there is very little discussion about how to protect these groups from stigma or harm.
The fourth theme concerns fairness. The computer programs used for genetic diagnosis are trained on data that mostly comes from people of European ancestry. This means the tools may not work as well for people from other backgrounds, yet there are few studies that measure these errors specifically for different populations. The fifth theme is about consent. People usually sign a one-time permission slip to have their DNA tested, but genetic knowledge changes over time. As computers learn new things, they might re-analyze old data and find new risks. The study points out that we do not have good systems for asking for permission again when these new findings appear. The final theme is about rules and laws. Governments are creating new regulations, but they often treat artificial intelligence and genetic data as separate issues. The study argues that we need rules that understand how these two fields work together.
The researchers conclude that the field is currently missing a framework that is built specifically for the unique nature of genetic AI. They argue that we cannot simply take rules from general medicine or general genetics and paste them onto this new technology. Instead, we need a new approach that recognizes how responsibility, privacy, and fairness work differently when a machine is interpreting our family history. This study serves as a map, showing exactly where the gaps are so that future research and laws can fill them. The authors plan to follow up with two more papers that will propose a new way of thinking about responsibility and apply these ideas to specific guidelines being developed in China. By identifying the blind spots in current thinking, this work aims to ensure that as we use artificial intelligence to read our genetic code, we do so in a way that protects individuals, families, and communities.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.