AI Scans Lungs, Assists Brain Surgery. Cancer Remains Unimpressed. The Biology Is Harder Than The Press Release.
AI is deployed clinically for lung image analysis and brain tumour removal assistance, while also being applied to drug development and treatment planning. Researchers caution that the complexity of cancer is poorly understood, and AI alone cannot defeat it. The technology does offer clinicians hope for reducing administrative burden and enabling more personalised care. The Guardian's interactive piece surveys the gap between tech industry promises and oncological reality.
This illustrates what I call the Problem Framing Fallacy. AI excels at narrow, well-bounded tasks like image classification. It struggles with systems that have irreducible biological complexity. The lesson is that computing power does not equal scientific understanding. You cannot model what you do not comprehend.
Researchers and clinicians working at the intersection of AI and oncology, as surveyed by The Guardian. The piece cites clinical deployment in lung imaging and brain tumour procedures, alongside drug development efforts.
- Open Google Scholar and search for 'AI cancer detection accuracy.' Read one abstract. Note the specificity of what the model actually detects versus the headline claim.
- Ask ChatGPT to explain the difference between classifying a tumour in an image and understanding tumourigenesis. The gap in its answer is the gap in the field.
- Pick one cancer type. Search for 'AI [cancer type] clinical trial 2025.' Observe how many are early phase. That is where the science actually sits.