Adam Smith Institute

View Original

It's almost as if they've no clue what they're doing

Two months ago we had this:

The transfer of 1.6 million patient records from an NHS trust to Google’s artificial intelligence subsidiary was legally inappropriate, the Department of Health’s data guardian has said.

Google’s Deep Mind division and the Royal Free trust in north London agreed two years ago to collaborate in developing an app to diagnose acute kidney injuries in NHS patients. It works by immediately reviewing blood test results for signs of deterioration.

To develop the Streams app, the trust made available to Google patient data going back five years, including sensitive details of patients who had not been treated or tested for kidney injuries. 

Using NHS data to train an AI to make the NHS work better is terrible. Those damn yankees and their technical skills, who do the bastard septics think they are trying to improve that Wonder of the World that is our very own National Health Service? They'll be forcing the ICU patients to pay $50 for an aspirin soon enough what with their only for money health care systems!

Today:

NHS patients will receive pioneering treatments assessed by an artificial intelligence program under plans being developed by the government’s medicines regulator, The Times has learnt.

The system, a prototype of which will come online next year, is expected to markedly speed up the appraisal of cancer drugs and other therapies from months to a matter of weeks.

Ooooh! Using NHS data to train an AI to make the NHS work better is just lovely.

Scientists at Microsoft, University College London (UCL) and Cochrane, a non-profit group that compiles reports on medical evidence, are developing an AI that can weigh up new health studies for itself. 

Yea, even if the septics are involved.

It's as if the government doesn't know what it's doing, isn't it? And that cannot possibly be true about the NHS, an absurd thought. For we all know that it's so exquisitely managed that no system in the world comes near it.