I think you confused my statement of caring what other people think about my suit, to wearing a suit. I wear suits. I just don't care what the fashion around them is.
> Once you distinguish all events you’re fine with and want to migrate your relational data, don’t try to cheat; don’t put your events as small and granular. Relational data is flattened; if you try to retrofit what happened from the final state, you will likely fail or not be precise at best.
The article is aimed at people who already have relational data, and want to build an event-driven system (whose events will eventually end up as relational data again downstream.)
Your system A might look like:
| Name | Balance |
| Michael | $3.03 |
You might design your system B to have the events AmountCredited{name, amount} and AmountDebited{name, amount}.
You don't know how system A ended up at its current state. That's what's meant by "flattened". When you want to "migrate the relational data", i.e. convert system A's relations into events, it's tempting to use the obvious AmountCredited{"Michael", $3.03} because you know it will result in Michael having the correct balance in the final system.
But it's not good to reuse AmountCredited, _because no such event actually happened_, which is why it could be called "cheating". If future-you looks at historical data, $3.03 won't correspond to any real transaction that happened. Instead you should instead make a special event like AmountImported{name, amount}.
For this example the convention in accounting is to use 'balance brought forward'.
It's a real transaction that happened, and everyone knows what it means i.e. the previous ledger with this account was closed off and the new one has the balance that was there when th book was closed.
Using 'imported' describes what you did, but not what the intention was.
Here’s a recent (yesterday) example of a benefit though.
I tried unsuccessfully to search for an ECG analysis term (EAR or EA Run) using Google, DDG, etc. There was no magic set of quoting, search terms, etc. that could explain what those terms were. Ear is just too common for a word.
ChatGPT however was able to take the context of the question I had (an ECG analysis) and lead me to the answer right away of what EAR meant.
I wasn’t seeking medical advice though, just a better search engine with context. So there are clearly benefits here too.
Yeah I took the person's comment, snipped "ECG analysis term EAR meaning" from it and popped that in to google, found a page "ECG Glossary, Terms, & Terminology List" and under the E section it has "Ectopic Atrial Rhythm".
That said if its correct, LLMs are less work for this kind of thing.
But often when people explain to another person what they are looking for, as done in their comment, they will do a better job explaining what they need than when they are in their own head trying to google it. Which is why I just snipped their words to search for it.
I used to work on a healthcare AI chatbot startup before traditional LLMs like BERT. We were definitely worried about accuracy and reliability of the medical advice then, and we had clinicians working closely to make sure the dialog trees were trustworthy. I work in aerospace medicine and aviation safety now, and I constantly encounter inadvisable use of LLMs and a lack of effective evaluation methods (especially for domain-specific LLMs).
I appreciate the advisory notice in the README and the recommendation against using this in settings that may impact people. I sincerely hope that it's used ethically and responsibly.
Maybe it’s ok to worry about both? Not trusting ”arbitrary thing A” does not logically make ”arbitrary thing B” more trustworthy. I do realise that these models intend to (incrementally) represent collective knowledge and may get there in the future. But if you worry about A, why not worry about B which is based on A?
You seem to be assuming, without any evidence at all, that LLMs giving medical advice are likely to be roughly equivalent in accuracy to doctors who are actually examining the patient and not just processing language, just because you are aware that medical mistakes are common.
"Six patients 65 years or older (2 women and 4 men) were included in the analysis. The accuracy of the primary diagnoses made by GPT-4, clinicians, and Isabel DDx Companion was 4 of 6 patients (66.7%), 2 of 6 patients (33.3%), and 0 patients, respectively. If including differential diagnoses, the accuracy was 5 of 6 (83.3%) for GPT-4, 3 of 6 (50.0%) for clinicians, and 2 of 6 (33.3%) for Isabel DDx Companion"
Six patients is a long way from persuasive evidence, because with so few patients randomness is going to be a large factor. And it appears that the six were chosen from the set of patients that doctors were having trouble diagnosing, which may put a thumb on the scale against doctors. But yes, it certainly suggests that a larger study might be worth doing (also including patients diagnosed correctly by doctors, to catch cases where GPT-4 doesn't do as well).
It's not whataboutism at its best, no. Just as with self-driving cars, medical AIs don't have to be perfect, or even to cause zero deaths. They just have to improve the current situation.
It depends who the end user is. As an aid for a trained physician, who is in a better position to spot the hallucinations, it may be fine, whereas a self-medicating patient could be at risk.
We absolutely need more resources in healthcare throughout the world, and it may be that these models, or even AGI, have great potential as a companion for e.g. Doctors Without Borders or even at the local hospital in the future. But there’s quite a bit more nuance to giving medical advice compared to perfecting a self driving car.
A self driving car can cause incredible damage straight away. I don't think you should underestimate that. But we also don't have enough healthcare access, so the need is more urgent than that for automated drivers, the health benefit of which is often only about reducing risk of driving while tired or intoxicated.
Yes a patient could be at risk - they're at risk from everything, including a poorly trained/outdated doctor. And even more at risk from just not having access to a doctor. That's the point: it's a risk on both sides; weighing competing risks is not whataboutism.
I am personally excited for the possibilities. Nobody should be using a LLM without verifying. Will some people do it? Of course, I remember that court case where the lawyer used ChatGPT and it made up cases.
If someone is going to make that mistake, there were other mistakes happening, not just using a LLM.
On the positive note. LLMs offer the chance to potentially do much better diagnosing on hard to figure out cases.
On the other hand, your MD is going to look for the obvious, or statistically relevant, or currently prominent disease.
But they could be presented 99% probability for flu, 1% or wazalla, and that testing for wazalla means pinching your ear tout may actually be correctly diagnosed sometimes.
It is not that MDs are incompetent, it is just that when wazalla was briefly mentioned during their studies, they happened to be in the toilets and missed it. Flu was mentioned 76 times because it is common.
Disclaimer: I know medicine from "House, MD" but also witnessed a miraculous diagnosis on my father just because his MD happened to read an obscure article
(for the story, he was diagnosed with a worm-induced illness that happened one or twice a year in France in the 80's. The worm was from a beach in Brazil, and my dad never travelled to Americas. He was kindly asked to provide a sample of blood to help research in France, which he did. Finally the drug to heal him was available in one pharmacy in Paris and in Lyon. We expected a hefty cost (though it is all covered in France), it costed 5 franks or so. But we were told with my brother to keep an eye on him as he may become delusional and try to jump through the window. The poor man cold hardly blink before we were on him:)
Ah, and the pills were 2cm wide, looked like they were for an elephant. And he had 5 or so to swallow)
You heard about the bell curve concept? Chances are about half the doctors you see are at the lower part of the curve. Which means they are borderline or completely incompetent at what they do.
I'd take my chances with a "properly trained" AI any day. Problem is, most medical corpus is full of bogus studies that have never been replicated, so it might be close to junk at this stage.
> It will lead to deaths,
regular doctors kill people everyday and get away with it because you accept the risks. What's different?
It's true. Only people like me should be allowed access to LLMs. Folks like you should be protected. Equivalent to accredited investor, there should be a tier of "knowledgeable normal person" who is allowed to do whatever.
That's a fair point. Most programmers are terrible. As a lead programmer I've dealt with countless "programmers" that can not do their job without me spoon-feeding them code.
There are lots of reasons why offices tend towards bad rather than good. If I sat and thought about it I could probably list a dozen systematic biases towards crappy office spaces. Here are just a few off the top of my head:
- Open-plan offices, hot-desking, and other negative patterns are more cost-effective for a given amount of space.
- Cookie-cutter, one-size-fits-all spaces are "easier" to manage from a personnel/HR point of view. Less griping about who gets the "good office," as there is no good office. No need to think about differences between people, we can just treat them as fungible "human resources."
- Similar race to the bottom regarding amenities (coffee, break rooms, etc.). Just look in the other comments of this thread. It's easier to target the least common denominator than provide personalized/individualized benefits in a manner that's fair to everyone.
- Insecure, distrusting managers promote bad office spaces like open plans, hot desking, etc. in order to better micro-manage their teams. Good managers can push back, but in practice the bad managers tend to be the squeaky wheels and get the grease.
I've worked for 10 different companies so far in my career. All but one I would consider a good company. But of those nine "good" companies, I have had one good office space. That's why I've been remote for the last 6 years. For me personally it was either go remote or leave the industry. I'm never making a open-plan or even cube-farm layout my primary working space again.
This isn't restricted to "offices." It's also why it's difficult to live with other people and housemates often end up in conflict. The more people you need to share a space with, the more contention each shared resource gets, and fewer individuals get what they actually want when they want it most of the time. There's tradeoffs to make, of course. We have societies and communities precisely because groups tend to accomplish more than individuals thanks to specialization and division of labor, but there's a reason humans tend to live in roughly family-sized units when given the choice, not open dorms with hundreds of other people.
Says anyone who has spent more than a decade in the workforce.
> There's no more reason for an office space to be bad rather than good. And there's no reason to think that having a good one is based on luck.
By “no reason”, you mean no logical reason. That might be true. Unfortunately a crafted reality controlled by sociopaths isn’t required to be logically consistent at every layer. Hence our current reality, where this is very much a case of luck.
> If office space dynamics and setup are important to you, then put it in your job search mental list and find a job that matches that.
Thanks for the advice. That’s /exactly/ what all the folks working remotely have done. That’s also why we are agitated by shitty management trying to take it away.
Yea I can’t help but realize after my career in offices then seeing mass wfh, the sociopath layer depended deeply on a perceived panopticon of working in the office. The panopticon was a negative motivator for workers, but also a common experience that standardized the culture and expectations.
WFH was like setting rats in a maze to free range and noticing they can be more productive but at the expense of common purpose. This brings a new dimension to the notion of “productive” that the sociopath layer is uncomfortable with. I think because it implies workforce instability.
Even Google, a company that purported to be about worker freedom to harvest productivity if top workers, has retreated to this position.
It’s a bit shocking to me still nearly 4 years later.
> Even Google, a company that purported to be about worker freedom to harvest productivity if top workers, has retreated to this position.
If you take the approach of "watch what they do, not what they say", Google is one of the clearest examples of pro-RTO: they invest in so many perks because they want to make the panopticon feel comfy.
There are other companies that don't give you a darn thing and just dangle the loss of one's job as a threat (Amazon), but I think top performers are more swayed by the "free food" approach than the "let's have a big public dashboard with the entire team's attendance" approach (again, Amazon).
> Says anyone who has spent more than a decade in the workforce.
Well, no. I have worked almost double that, and I don't feel that way.
I tend to think there is mainly an echo chamber of somewhat entitled young north americans incorrectly correlating high compensation in a decade long hyped industry (tech) with social status.
> a crafted reality controlled by sociopaths
Well if that is your definition of reality, maybe you should consider whether you could be part of your issue with office spaces...
> shitty management trying to take it away
Definitely, that kind of remark reinforces my opinion that you should reflect on whether your look at the situation is biased.
Anecdotally, I am aware of people who seem to truly believe that. I used to think it is a question of age, but even that quickly got corrected as from within my professional circle there was no clear way to determine a good predictor for office/no office preference. It seems oddly almost evenly split, which seems very weird to me.
That said, the 'entitled young Americans' thing appears to be a common talking point I see on linkedin and other corporate safe spaces.
"Have you considered the increased opportunities for synergies in the office? We should punt on this until next quarter's all hands when we can utilize our increased office presence to actualize our OKRs."
I agree in general that there needs to be at least some level of review (typos, etc.), not necesssarily to catch big but subtle issues.
On the other hand, I've had it where small (single character!) PRs have to wait for several days because I need to keep bothering my team to review them. But this seems like a problem that solvable organizationally somehow (not sure how?) rather than by eliminating an important part of the process.
Yeah, I know that I tend to avoid the broken stuff in my own code and/or only test it along happy paths because I know how it's supposed to work. I can work around that, sure, but IME there's a certain kind of blindness to things that comes with familiarity - a fresh set of eyes is a good way to mitigate that.
I used to keep a stock of $2 bills around specifically to pay bus fare - with the bill and a couple quarters in hand, I could step onto the bus, pay, and move back to a seat with barely a pause. This sometimes confused the drivers.
Must be a frustrating, expensive, kinda sad way to live life.