Who takes responsibility when AI gets a diagnosis wrong? Dubai health chief explains

Date:

DHA chief says accountability for AI errors will depend on who uses the technology and how it is applied.

Dubai: When an artificial intelligence (AI) system contributes to a wrong diagnosis, responsibility does not automatically fall on one party. It depends on how the technology is used, according to Dr Alawi Alsheikh-Ali, Director General of Dubai Health Authority (DHA).

Speaking at the AI Retreat during Dubai Future Week, in a session titled ‘Past the Pilot’, Alsheikh-Ali addressed a question that many patients may have: If AI leads to a misdiagnosis, who is held responsible?

He identified three potential parties: the physician, the hospital that approved the technology and the company that developed it.

He then distinguished between two scenarios: AI acting as a tool to assist a doctor, and AI functioning as the healthcare provider itself. On the latter, he said: “I don’t think we’re there yet.”

When AI is used to assist a physician, the doctor remains responsible for overseeing the process and making the final decision. However, if the error stems from a defect in the technology, the developer may share liability. Hospitals could also bear some responsibility if they allow staff to use AI without proper evaluation, monitoring or training.

“It’s not a straightforward answer,” he said, adding that responsibility would probably be shared.

Patients should know AI’s role

Alsheikh-Ali stressed the importance of patients knowing when AI is involved in their diagnosis or treatment. He illustrated the challenges through examples from clinical practice.

One example involved a patient visiting a clinic with persistent chest pain. When a doctor asks whether he is experiencing chest pain, the patient may answer no and genuinely mean it. In another scenario, he may say no while his expression suggests otherwise. In a third, his wife may intervene, saying: “I can’t believe this. He’s in denial again.”

An ambient AI tool recording the consultation might transcribe the same word in all three situations, failing to capture the context behind the response.

“In those three examples, the no is not the same,” Alsheikh-Ali said.

Such tools are already being used in the United States, raising questions about how AI can recognise and interpret the nuances of human communication and clinical judgement.

When does AI add value to patient care?

Alsheikh-Ali also highlighted the importance of weighing the benefits of AI against its potential drawbacks, citing an example involving cancer screening.

He described a technology capable of detecting a particular cancer six months earlier than existing screening methods.

However, further research found that the system also flagged findings that were not cancer, leading to additional tests and greater anxiety among patients. More importantly, detecting the cancer earlier did not change the final outcome for patients.

The example raised an important question: Does introducing AI actually improve patient care, or does it simply create additional costs, procedures and stress?

For Alsheikh-Ali, the focus should be on whether a technology delivers measurable benefits to patients, including saving time, reducing costs or improving health outcomes.

He cautioned against using AI simply as a marketing tool. Promoting a hospital as an institution that uses AI may create a modern image, but that alone does not translate into better patient care.

Instead, he said, authorities and healthcare providers must assess what needs to be established before allowing an AI system into clinical practice.

Lower-risk applications may require less stringent evaluation, while high-risk technologies should face more rigorous checks. Any new tool should demonstrate that it can make healthcare better, safer or more efficient, including helping providers treat more patients without compromising safety.

Where AI is already making a difference

According to Alsheikh-Ali, some of the most promising applications of AI in healthcare today are in hospital operations and pattern recognition.

These include summarising clinical notes and analysing medical images such as X-rays, CT scans and MRIs, as well as interpreting skin and heart readings.

He also pointed to the potential of AI in drug discovery, where the technology could help researchers identify new targets and shorten the time needed for scientific development.

However, one of AI’s biggest contributions to healthcare could be giving doctors more time with their patients.

Doctors spend considerable hours on paperwork and administrative tasks. If AI can take over some of that workload, Alsheikh-Ali said, it could allow physicians to focus more on patient care and restore the human connection at the heart of medicine.

Share post:

Subscribe

spot_imgspot_img

Popular

More like this
Related

UAE graduates first batch of community emergency response volunteers

Trained volunteers graduate from accredited programme to strengthen emergency...

GDRFA Dubai secures landmark accreditation for its business incubator

Dubai SME recognises GDRFA's Innovation and Future Centre for...

Palm Jumeirah VIP parking rates soar to Dh50 per hour

Motorists face Dh100 for two hours of parking, with...

Things to do in Dubai this weekend: November 20-22

From dog-friendly outings to hands-on cooking masterclasses, here are...