Artificial intelligence (AI) is transforming healthcare, and one of its most visible applications is making its way into the exam room.

A growing number of doctors are using AI-powered “ambient listening” tools to record and summarize patient visits, allowing them to focus more on patient care and less on administrative tasks.

According to the WSJ, this technology, which listens to conversations between doctors and patients and generates clinical summaries, is expanding rapidly in hospitals and emergency rooms across the United States.

However, while it promises significant benefits, it also raises concerns about privacy, accuracy, and the patient-doctor relationship.

Ambient listening technology, such as the DAX Copilot app used by many healthcare providers, records conversations during medical appointments with patient consent.

The AI processes the dialogue, filters out non-clinical information (like casual chit-chat about a patient’s family or pets), and produces a structured clinical note that the doctor can review, edit, and upload to the patient’s medical record.

This process typically takes just a few minutes, a stark contrast to the hours doctors traditionally spend on documentation.

For example, pediatrician Dr. Jocelyn Wilson, as reported by the Wall Street Journal, uses DAX Copilot to save over an hour daily on entering appointment details, allowing her to maintain eye contact with patients instead of typing during visits.

The technology is designed to alleviate the administrative burden that contributes to physician burnout.

A 2020 Mayo Clinic study found that doctors spend one to two hours outside of office hours writing notes, cutting into personal and family time.

By automating this process, tools like DAX Copilot have reduced documentation time by 24% in some cases, enabling physicians to finish clinics on time and spend less time charting at night.

Dr. Nishit Patel, a dermatologist and Chief Medical Informatics Officer at Tampa General Hospital, noted that for every hour spent with patients, doctors typically spend two hours on documentation.

Ambient AI allows them to review and tweak AI-generated notes rather than starting from scratch, significantly improving efficiency.

The primary advantage of ambient listening technology is its ability to free doctors from the keyboard, fostering more meaningful patient interactions.

Dr. Eric Poon, chief health information officer at Duke Medicine, shared that since adopting DAX Copilot, he has been able to step back from taking telegraphic notes and engage in more natural conversations with patients.

This shift not only improves the quality of the patient-doctor relationship but also helps doctors capture details they might otherwise miss during hectic appointments.

For instance, Dr. Woo, as cited in other sources, noted that Abridge, another AI tool, caught critical patient information she overlooked, leading to updates in treatment plans.

Patients also benefit from this technology.

A patient named Tabatha Conza, treated by Dr. Patel, appreciated the increased eye contact and one-on-one engagement during her visit, describing it as a more personal experience compared to doctors distracted by computers.

A 2024 survey indicated that about 70% of patients nationally are comfortable with physicians using AI in appointments, suggesting broad acceptance when the technology is explained transparently.

Moreover, the technology can lead to cost savings.

A 2019 Mayo Clinic study estimated that emergency room doctors without assistants spend over two hours charting per shift, costing up to $600 daily.

By reducing this time, AI tools like Abridge and DAX Copilot can lower operational costs while improving care delivery.

Despite its advantages, ambient listening technology raises significant concerns about privacy and accuracy.

Patients worry about where recordings are stored and whether they could be hacked or misused.

Atrium Health, which uses DAX Copilot, ensures that recordings are accessible only through biometrics or password authentication and are deleted after doctors approve the clinical notes.

Similarly, Tampa General Hospital uploads audio files to secure servers, not storing them on doctors’ phones, and complies with Florida’s two-party consent law by informing patients and offering an opt-out option.

However, in states with one-party consent laws, it’s unclear whether all facilities notify patients, potentially leading to ethical concerns.

Accuracy is another issue. AI systems can produce “hallucinations,” where they generate incorrect or fabricated information.

Research by Stanford University found that while AI-generated summaries of radiology scans were often preferred over human-written ones, they could miss nuances related to a patient’s age, gender, or cultural background.

Dr. LaTasha Seliby Perkins, a family physician, emphasized the need for doctors to review AI outputs carefully to avoid errors, particularly since speech-to-text tools may have higher error rates for certain demographics, such as Black speakers.

Some patients express unease about AI’s presence in the exam room. While many, like Conza, are comfortable with it, others perceive it as intrusive or “Big Brother-like.”

A 2024 survey found that over half of patients consider healthcare AI “a little scary,” with 70% expressing data privacy concerns. These fears are compounded by the potential for AI systems to be trained on patient data, raising questions about consent and data security.