People love to bring this up, and it was a silly thing to say -- particularly since he didn't seem to understand that radiologists only spend a small part of their time reading scans.
But he said it in the context of a Q&A session that happened to be recorded. Unless you're a skilled politician who can give answers without actually saying anything, you're going to say silly things once in a while in unscripted settings.
Besides that, I'd hardly call Geoffrey Hinton an AI evangelist. He's more on the AI doomer side of the fence.
No, this was not an off-hand remark. He made a whole story comparing the profession to the coyote from road runner “they’ve already run of the cliff but don’t even realize it”. It was callous, and showed a total ignorance of the fact that medicine might be more than pixel classification.
Radiologists, here, mostly sit at home, read scan and dictate reports. They rarely talk to other doctors and talking to a patient is beyond them. They are some of the specialists with the best salary.
With interventional radiologists and radio-oncologists it's different but were talking about radiologists here...
I'm a radiologist and spend 50% of my time either talking to patients or other clinicians.
You practice in Québec ? If so I am quite surprised, because my wife had a lot of scans and we never met a radiologists who wasn't a radio-oncologist. And her oncologist never talked with the radiologists either. The communication between them was always through written demands and reports. And the situation is similar between her neurologist and the radiologists.
By the way, even if I sound dismissive I have great respect for the skills required by your profession. Reading an IRM is really hard when you have the radiologist report in hand and to my untrained eyes it's impossible without it!
And since you talk to patients frequently, I have an even greater respect of you as a radiologist.
My wife’s an ER doctor and she talks to radiologists all the time.
I also recently had surgery and the surgeon consulted with the radiologist that read my MRI before operating.
Then it's an organizational problem (or choice) in the specific hospital where my wife is treated/followed and I apologize to all radiologists that actually talk to peoples in a professional capacity!
Or maybe it's related to socialized Healthcare because in the article there is a breakdown of the time spent by a radiologists in Vancouver and talking to patients isn't part of it.
I would argue an "AI doomer" is a negatively charged type of evangelist. What the doomer and the positive evangelist have in common is a massive overestimation of (current-gen) AI's capabilities.
Many of us have changed opinions after seeing how the tech does not scale.
At the time? I would say he was a AI evangelist.
The tech scales, but accessing the training data is a real problem. It's not like scraping the whole internet. And most of it is unlabeled.
I think in general this lack affects almost all areas of human endeavor. All my speech teaching my kids how to think clearly, to young software engineers about how to build software in a some giant ass bureaucracy, how to debug some tricky problem, none of that sort of discovering truth one step at a time or teaching new stuff is in blogs or anything outside the moment.
When I do write something up, it is usually very finalized at that time; the process of getting to that point is not recorded.
The models maybe need more naturalistic data and more data from working things out.
If you need more data to scale, and there is no data, it literally can't scale.
Scale is not always about trougput. You can be constrained by many things, in this case, data.
I was unclear. There is data, but it is expensive to access, so the value proposition is often not there without some beneficent entity.
It's the power of confidence and credentials in action. Which is why you should, when possible, look at the underlying logic and not just the conclusion derived from it. As this catches a lot of fluff that would otherwise be Trojan-Horsed into your worldview.
[dead]