Researchers will gather at Dartmouth College on Wednesday to discuss the promise AI holds in health care – as well as the challenges it presents.

The symposium, AI in Medicine: Bridging Innovation & Practice, is open to the public as well as academics.

“We believe that if AI technologies and these innovations are used rigorously and responsibly, they have the potential to benefit all of us, especially disadvantaged and under-resourced populations,” said Saeed Hassanpour, who leads Dartmouth’s Center for Precision Health.

We’re at a “turning point” when it comes to AI in medicine, said Hassanpour, who’s also professor in Dartmouth’s epidemiology, computer science and biomedical data science departments.

Artificial intelligence technology has advanced at the same time that modern healthcare has become an increasingly rich trove of data, with electronic medical records, genetic testing and other digital tools. He said that’s why AI could potentially reshape the field.

The technology can sift through vast amounts of data, detecting patterns that are hard for humans to see – such as helping radiologists spot subtle signs of cancer with greater accuracy, or informing which course of treatment is most likely to work for a patient with a certain medical history.

But Hassanpour said the spread of AI in medicine also carries risks, if it’s not developed thoughtfully. AI is only as good as the data it’s trained on – and if that data reflects existing biases or excludes certain demographic groups, it could end up reinforcing health disparities. Hassanpour said it’s important to have those conversations now. Otherwise, he said, “we might end up with technology that is not aligned with our values, benefits only a few, and exacerbates biases and disparities in our society.”