Despite AI, burnout continues to dog radiologists

It’s been at least 10 years since AI started promising to save radiologists stress and time so they could unhurriedly consult with referring physicians and even discuss results directly with patients. 

After all, wasn’t radiology the fastest medical specialty to embrace and adopt AI? And yet, in 2025—nearly 10 years since the technology set the profession abuzz—burnout continues to afflict many if not most radiologists. 

In fact, it’s the No. 1 complaint of radiology professionals around the world. 

That’s according to a global survey of 708 radiologists from 50 countries conducted by the U.K. teleradiology practice Everlight Radiology. 

The survey also found AI very much on the minds of respondents. Here are eight key findings from that section.

1. Many radiologists are not regularly using AI.

More than half the responding radiologists, 57%, indicated they do not use AI routinely in their roles. The authors comment: 

‘AI is largely seen as a valuable tool to support, not replace, human expertise.’

2. AI adoption rates vary considerably by region.

Everlight found the highest adoption in Australia, where 67% of radiologists report using AI. Runners-up are the U.K. at 61% and New Zealand at 48%.

‘Ireland has the lowest rate of AI adoption—less than a quarter (24%).’

3. A majority of radiologists (57%) do not use AI routinely in the course of their work.

‘Of those who do (37%), its primary function is to assist in detecting subtle findings or to provide a secondary opinion (27%).’

4. In the U.K., NHS employees have high awareness of AI adoption in their workplaces—but most do not use it routinely. 

Among those who do, NHS staff are more likely to report finding AI unhelpful (8%).

‘That stands in contrast to lower levels of dissatisfaction reported by non-UK public hospitals (3%) and private radiology providers (3%).’

5. Radiologists hold mixed views on the relative importance of AI implementation. 

While 56% of respondents believe their employer should prioritize investments in AI, enthusiasm varies by geography.

‘The highest support levels show up in Dubai (86%), Australia (70%) and New Zealand (64%). Support levels are notably lower in Ireland (54%) and the U.K. (53%).’ 

6. Despite the international variations, there is a general consensus on AI’s valuable role in streamlining nonclinical tasks.

Some 83% recognize AI’s utility in administrative workflows, and 85% believe it enhances productivity and efficiency. Yet an overwhelming majority (95%) also acknowledge AI’s inherent limitations and risk of error, reinforcing that radiologists themselves remain indispensable. 

‘Indeed, 81% of respondents agree strongly that AI cannot replace the specialized skills and training of radiologists.’ 

7. When asked about AI’s effectiveness in reducing diagnostic errors compared to peer review, opinions diverged.

While 38% believe AI can surpass peer review in error reduction, a larger proportion (48%) disagree.

‘This reveals a split in opinions on AI’s role and constraints within radiology.’

8. A large majority of radiologists (78%) view AI as a support tool or ‘second set of eyes’ rather than a primary solution in diagnostic work.

‘Although 73% recognize the benefits of AI in enhancing radiological practice, 19% see no value in its application.’ 

As the Everlight analysts see it, these findings suggest that, while AI holds promise, “there remains some cynicism around its clinical merits, and radiologists tend to see it as one part of a larger, multifaceted approach to addressing the field’s challenges”—including rampant burnout—“rather than a cure-all solution.”

The report is available in full for free.