So, the Theory of Everything won an Oscar yesterday. As a result, I had a great discussion with Dr Malcolm Ireland today (Director of Training at Wentwest). We were discussing Stephen Hawking's recent comment that artificial intelligence will be 'the end of mankind'.
Malcolm made two great points:
1. Expert systems have historically worked deductively... start with a large amount of information, then try to reduce the number of possibilities through iterative questioning until you arrive at the likely answer. And yet expert humans don't do that. We don't really know how they do it... some call it pattern recognition, or intuition... but a great clinician works in reverse. "A 67 year old man presents with cough, haemoptysis..." and they are already deducing that it may be TB, Ca Lung, acute pneumonia. They then work backwords, hypothesis testing against the likely outcomes, rather than starting with an exhaustive list and testing each one. Maybe that is what it is to master an art. It may be a while before machines are there (but they are getting smarter every day!!)
2. Malcolm also thought that, whilst machines rely on large data stores, so much of that data has been created by humans, is incomplete, is stored in inaccessible places, is not interoperable and is, essentially, human and messy. As a result, machines will have trouble for quite some time becoming highly accurate in some of the 'greyer' areas of human knowledge. In effect, our own human messiness and incompetence may save us yet!!
What do you think? Are the machines coming?...
Stephen is a GP Supervisor, Medical Educator, GP academic and Medical Director of Medcast. He has completed a PhD on Virtual Communities of Practice in GP Training.
Become a member and get unlimited access to 100s of hours of premium education.
Learn moreThis article explores the risks and management of dynamic hyperinflation in mechanically ventilated patients with status asthmaticus. It outlines causes such as airway obstruction and bronchospasm, and provides strategies for adjusting ventilation settings to minimise gas trapping, support cardiac function, and reduce complications like pneumothorax and systemic hypotension.
This is the second FastTrack on atrial fibrillation (AF). Part 1 described the diagnosis, initial assessment of AF, and indications for urgent review. This FastTrack describes the approach to setting a management strategy using risk calculators, how to assess and reduce risk factors, and long term monitoring of patients with AF.
While cardiac arrest is often a rare event, the high stakes nature means we need to be able to respond instinctively, ensuring immediate effective action when it matters most. How do you stay prepared in-between ALS or PALS courses?