The metric of radiology jobs as a sign of the lack of AI revolution in the field seems poor to me. Sadly, much of our medical infrastructure (and the jobs it creates) only have a very tenuous relationship to the actual care and the quality that it delivers. Rather, most of the infrastructure tries to optimize for the billing and the legislation that surrounds it.
One of the ways to immediately see this is for us, a technical crowd, to puzzle on why it seems that Moore's law doesn't seem to affect medical technology... AT ALL. Some of the same procedures using the same machines from decades ago today cost A LOT more than they used to, for instance.
This isn't to say that this AI revolution in radiology hasn't been underwhelming; I just think that using this job metric is a poor indicator of the technology's capability.
One of the ways to immediately see this is for us, a technical crowd, to puzzle on why it seems that Moore's law doesn't seem to affect medical technology... AT ALL. Some of the same procedures using the same machines from decades ago today cost A LOT more than they used to, for instance.
This isn't to say that this AI revolution in radiology hasn't been underwhelming; I just think that using this job metric is a poor indicator of the technology's capability.