This is IMO, the single biggest thing that mithers me about the TBI “AI will fix the NHS” stuff. We have decades of really robust evidence from across the world that technological innovation does not lower health costs now, quite the reverse!

Tech companies want to win in AI. Winning in AI requires access to large datasets to train on. Governments have huge, interesting datasets. UK has the *best* health datasets on the planet. Tech companies pay very clever lobbyists to get access to datasets. What happens next, who can say? 🙄

My mam was working in DSS/DWP in the 90s, when that got 'computerised'. That certainly didn't result in reduced costs or less work for her and her colleagues. I sometimes wonder if that's cause it was managed so badly - the bosses assumed costs could be cut *immediately*.

Bevan to Cripps?

It'll save some money in some contexts. It's likely that the NHS will come to spend less on e.g. locum radiologists as AI increases productivity in clinical radiology. This does however just mean that people are diagnosed faster, have better survival rates, and so come to cost more in the long-run.

My standard answer to "can AI save the state loads of money" is "did the internet save the state loads of money?"My standard answer to "can AI save the state loads of money" is "did the internet save the state loads of money?"

Yes, also NHS has terrible IT system (which on its own causes many NHS problems) so how is AI going to work in that environment, even if you think AI can fix the NHS (which i don't, it can do certain things but like you said won't lower costs because will increase what can be done).

I suspect the value of AI to the NHS is almost entirely different to what most folks think it will be tho (I think much more promising that it can drive pharma costs down and that'll end up looking more like more impressive medicines more than the organization itself being cheaper)

What I’ve seen is a lot of stuff being labelled “AI” that actually amounts to “computerised staff rostering” - with pretty limited intelligence! There’s great claims for scan readings but picking up minute tumours isn’t necessarily of benefit to anyone - read Margaret McCartney’s The Patient Paradox

There are strong incentives to do the expensive thing rather than the low cost thing. See current excited investment in lung cancer vaccine at goodness knows what cost versus the boring low cost smoking cessation services which would sort out most of the lung cancer (+ CVD etc)

And the effects are extreme now - when that Yes Minister sketch on smoking debuted, the cost of dementia care etc was a fraction of current costs