Want to know how artificial intelligence will change your job? Let’s take a lesson from the Department of Radiology. Radiology is now a hot topic in the artificial intelligence race. The field was mentioned several times by tech executives at the World Economic Forum in Davos last month and was discussed in a White House white paper on artificial intelligence and the economy.
A radiologist examines a mammogram on May 6, 2010, in Los Angeles.
Radiology is not the only profession affected by AI, as the technology is being integrated into the work of many professions including software engineers, teachers, and even plumbers. Goldman Sachs Group estimates that if AI-related technology is widely used, it could replace 6% to 7% of the U.S. workforce, but the technology is also expected to create new jobs.
But the field of radiology has become a prime example of how artificial intelligence can empower rather than replace jobs. Dr. Pak-ho Chen, a diagnostic radiology specialist at Cleveland Clinic, said the type of work in the radiology department is also well suited to be assisted by artificial intelligence.
The Department of Radiology has a large amount of data that can be used for artificial intelligence research and application, and the training of artificial intelligence requires massive data support. Artificial intelligence can process huge amounts of data far faster than humans, and is already helping to speed up some workflows in radiology departments – for example, determining which image scans need to be processed first.
In the radiology department, human doctors still need to perform most of the core tasks, such as making diagnoses, performing physical examinations on patients, and writing diagnostic reports. Moreover, as radiology departments continue to embrace artificial intelligence technology, job demand in this field is expected to grow faster than other industries.
Jack Kasten, a researcher at the Center for Security and Emerging Technologies at Georgetown University, said: "Not only has artificial intelligence not replaced radiology workers, it has actually improved their work processing capabilities and increased the demand for their professional services. For the technology industry, this can be said to be a good prospect for artificial intelligence to bring positive effects to the economy."
How artificial intelligence can empower work rather than replace it
Dr. Chen pointed out that artificial intelligence is good at analyzing images and identifying patterns in data, and these two abilities are key to the work of radiology departments. Additionally, radiology has been digitizing for many years, which means there is a huge amount of data available in the field.
He said: "There are still some niche work scenes that still use traditional analog methods, but in the United States, the vast majority of X-rays, CT, and MRI images have been stored digitally."
Dr. Chen and two other radiologists interviewed by CNN said that radiologists are now using artificial intelligence to prioritize image scans, improve image quality, and help write diagnostic report summaries.
Dr. Shadpur Demeri, an interventional radiologist at Johns Hopkins Medical Center, said: "Artificial intelligence has never replaced anyone's job, it has only made our work more efficient and valuable."
René Vidal, a professor of engineering and radiology at the University of Pennsylvania School of Engineering, believes that artificial intelligence will be particularly useful in obtaining high-quality MRI images with fewer scans. This technology can speed up the examination process and allow doctors to see more patients in the same time.
Vidal said that researchers are currently exploring other application scenarios of artificial intelligence, such as using it to measure tumor volume and automatically generate diagnostic reports, but it may be too early for these applications to be implemented.
Professions that were predicted to disappear but still exist
Vidal said AI tools used in the medical field must be approved by the U.S. Food and Drug Administration, a process that takes about eight years including the development and clinical trial stages. However, relevant approval work has been advancing: of the 1,357 artificial intelligence medical devices currently approved by the US Food and Drug Administration, 1,041 are suitable for use in the field of radiology.
At the same time, job demand in radiology departments continues to grow. The U.S. Bureau of Labor Statistics predicts that radiology jobs will grow 5 percent from 2024 to 2034, faster than the 3 percent average for all occupations. Data from recruitment platform Indeed obtained by CNN also shows that the number of radiology jobs will increase in 2025 compared with five years ago.
Radiology experts interviewed said that the rising demand for imaging examinations during medical diagnosis, coupled with the aging of the population, may be the main reasons for the increase in demand for radiology services.
But that’s not how the field used to be viewed. Nobel Prize-winning economist and computer scientist Jeffrey Hinton, known as the "father of artificial intelligence," said in 2016 that "it's time to stop training radiologists" because deep learning - a branch of artificial intelligence that simulates the learning patterns of the human brain - will be better at the job in 5 to 10 years.
Hinton said in an email to The New York Times last year that his 2016 comments were too absolute.
Demery recalls that around 2015 to 2016, there was widespread anxiety in radiology about artificial intelligence replacing human jobs. Today, this technology has been regarded as a “second pair of eyes” for doctors.
The hidden dangers of over-reliance on artificial intelligence
However, Dr. Chen also pointed out that artificial intelligence has the risk of bias, and people may also become over-reliant on it. For example, a 2022 MIT study showed that artificial intelligence can accurately determine a person's race from X-rays, unlike human radiologists, raising concerns about bias in diagnosis.
Dr. Chen also expressed his concern that if AI technology develops to a sufficiently mature stage, medical institutions may have the idea of adjustment of staffing - such as replacing doctors with nurses, or replacing radiology specialists with general practitioners. This approach may be feasible in some cases, but it is mostly not suitable for the detection of diseases that radiology departments mainly deal with, such as cancer and fatal infections.
He said: "We must understand that the good performance of artificial intelligence algorithms is largely due to the fact that its automated output results will be reviewed by professional doctors. It can be said that it is this collaboration between machines and professionals that truly improves the level of diagnosis and treatment."