A robot trained on videos of surgeries performed a lengthy phase of a gallbladder removal without human help. The robot operated for the first time on a lifelike patient, and during the operation, responded to and learned from voice commands from the team—like a novice surgeon working with a mentor.
The robot performed unflappably across trials and with the expertise of a skilled human surgeon, even during unexpected scenarios typical in real life medical emergencies.
At some point in a not very distant future, you will probably be better off with the robot/AI. As it will have wider knowledge of how to handle fringe cases than a human surgeon.
We are not there yet, but maybe in 10 years or maybe 20?
I’d bet on at least twenty years before it’s in general use, since this is a radical change and it makes sense to be cautious about new technology in medicine. Initial clinical trials for some common, simple surgeries within ten years, though.
This is one of those cases where an algorithm carefully trained on only relevant data can have value. It isn’t the same as feeding an LLM the unfiltered Internet and then expecting it to learn only from the non-crazy parts.
Hopefully more people learn that this is the important part.
It becomes nonsense when you just feed it everything and the kitchen sink. A well trained model works.
it’ll definitely get the greenlight in countries like China before anywhere in the west, I believe
Why?
Just a hunch, since technological advancements seem to hit the public realm much faster in places like China, in the cities especially. I don’t know what the laws are like there, but I’ve heard rumors that there is less government regulations for technologies that can benefit the general public, like drones and automated metros. Oh yeah, and how could I forget about the robots they show off at conventions, to take the place of receptionists and other customer-facing positions.
The main issue with any computer is that they can’t adapt to new situations. We can infer and work through new problems. The more variables the more “new” problems. The problem with biology is there isn’t really any hard set rules, there are almost always exceptions. The amount of functional memory and computing power is ridiculous for a computer. Driving works mostly because there are straightforward rules.
I doubt it. It simply would be enough, if the AI could understand and say when it reaches its limits and hand over to a human. But that is even hard for humans as Dunning & Kruger discovered.
Fringe cases yes, like rare conditions. It almost certainly won’t be able to handle something completely unexpected.
The AI will (probably) be familiar with every possible issue that no human will be able to match.
I’m not sure what kind of “completely unexpected” situation is possible can happen, that a normal surgeon would handle better?
But I agree it would have to be a lot smarter than current LLM and self driving for instance. Like a whole other level of smarter. But I think that is where we are heading.
I think you make a mistake of thinking that our collective body of knowledge is exhaustive. We discover new things all the time. Until we know everything (i.e. never), there will be gaps that AI will not be able to accommodate.
Would it be able to handle a sudden power outage? A fire alarm going off?
What happens to an ecmo machine during a power outage or fire alarm?
https://en.wikipedia.org/wiki/Extracorporeal_membrane_oxygenation
The idea should be to augment healthcare professionals with tools they can use. The hospital will need to have contingencies in place. I agree if that your point is that we can’t replace people with machines. But we can increase effectiveness with them.
As well as a human, and without fucking up because of stress.
Also my guess is these would be monitored by trained professionals.