An example of why we need to take things with a grain of salt...

Anthropic CEO Dario Amodei made waves with a claim that AI has fundamentally changed radiology, stating in an interview that 'the most highly technical part of the job has gone away,' suggesting radiologists now primarily discuss scans with patients. This assertion, coming from a respected AI leader, was quickly challenged by an MD/PhD candidate with direct clinical and research experience in the field. The expert called the claims 'baseless and completely incorrect,' highlighting a significant disconnect between AI hype and on-the-ground medical reality.

Contrary to Amodei's vision of a transformed field, the expert explains that radiology today uses numerous specialized, narrow AI models for tasks like counting lung nodules or detecting aneurysms. These tools are not a single generalized system and suffer from consistency issues, including false positives/negatives and failures with anatomical variations or post-surgical changes. Consequently, they have not meaningfully reduced radiologists' workload or replaced core technical duties. While some hospitals use AI for screening or prioritizing studies (like detecting intracranial bleeds), the field remains far from the paradigm shift described.

The viral critique serves as a crucial reminder of the 'incentives for companies to make exaggerated claims' and for researchers to overstate impact. It underscores that the most accurate assessment of AI's real-world utility often comes from domain experts, not tech executives. For professionals tracking AI advancements, this incident is a case study in applying healthy skepticism and seeking verification from practitioners within the target industry.