> Lack of JSON schema restriction is a significant barrier to entry on hooking LLMs up to a multi step process.
(Plug) I shipped a dedicated OpenAI-compatible API for this, jsonmode.com a couple weeks ago and just integrated Groq (they were nice enough to bump up the rate limits) so it's crazy fast. It's a WIP but so far very comparable to JSON output from frontier models, with some bonus features (web crawling etc).
Wait. Are you saying you don't get why some people may be bothered by sexual crimes against innocent children? Am I misinterpreting your post ?
Edit-- I'll happily take the down vote on this, I understand the policy of extending charitable interpretation But I am genuinely confused which is why I asked for clarification, it was not rhetorical.
Taking the "no moat" argument at face value, I think it's important to remember that some of the largest players in AI are lobbying for regulation too.
It is feasible actually. Graduated med school with two people in their late 30s, early 40s. You didn't ask for it but be forewarned: it'll take 10 years before you have a realistic taste of what the job actually is like. I don't usually speak in absolutes, but I will confidently claim that no amount of shadowing, training, volunteering in clinics/hospitals, or family will provide you with a shortcut to that experience. Could talk for days about this, but I have a cloud to go yell at.
I'm always in the middle of a test that I didn't study for. There's a nebulous higher stakes component to it beyond just failing. I can't read the questions for some reason, but I feel the topic of the test, and it's different every time. Some nights it's high school, others college, but now mostly licensure-related since I guess that was the last time I actually took a test.
It will reduce your risk. Poor sleep leads to an inflammatory cascade which can contribute to atherosclerosis. Also affects blood pressure; hypertension is a well-established risk factor for vascular disease. More reading here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3548567
> You are bringing it up because of the aura attached to it in in physics.
I didn't get that out of the original comment, and wouldn't presume to apply intent. I think most learned professionals here understand what is meant when they hear or say (for example) "that task took a lot of energy" or "I spent a lot of energy doing xyz..." without having to spell it out in academic terms.
Twin dad here. The struggle is real! This is just my 2c, but kids are designed to take and take until there's nothing left of you. It's not their fault, but that's just how it is. You can still be an engaged parent and have wonderful experiences with them, but you (major emphasis) cannot let your health suffer.
Keeping your mental sanity is as important as not letting them wander into a busy street or play with fire, etc. You are no good to them as an empty, stressed, beat down carcass of your former self. I have all kinds of strategies that I put into play when I find myself tapped out, but n=1 and I doubt it would be useful to others, I just wanted to share that I 100% get you. Take care of yourself, just like you would your kids.
Once upon a time I worked on a mobile app that was to be used by the VA to help with the management of medical trainees. I thought it was routine stuff (time logging, policy directories, etc)...
They actually wanted a system that would do that stuff too, but behind the scenes it would track the location of each user down to the room they were in. The administrators were suspicious that residents/trainees were sneaking off premises to "work from Starbucks" and they wanted ammunition to use against the sponsoring institutions to revoke payments (which go to resident salaries)... They weren't the slightest bit worried about the legal, moral, or safety implications. It never crossed their minds at all. (They never got what they wanted, at least not from me.)
Thanks for the feedback. I'll be thinking about a way to play around with it sans authentication. But just to clarify, you can try it out in the browser after getting a token. The sign up doesn't even have a captcha, and it's free, so while not 100% frictionless, it's pretty darn close.
The menu transparency resolves on mobile once you move down the page, but yeah, I appreciate the observation.
Good question about motivation: I was tasked with incorporating guidelines for a patient intake system, and decided to spin it out into a dedicated service. It's also being used in an EMR integration for their patient portal solution. And I use it every day in clinical practice myself, via postman.
OP here, happy to answer any questions. I'm sure there are some things I could have improved, but at a certain point it's better to push it out into the wild. All comments, suggestions, criticisms are welcome.
In reference to the psychological comorbidity of dealing with this issue, there are psychiatrists who are now specializing in post shock depression and anxiety. Not uncommon for cardiologists to refer all of their device patients to a mental health provider.
Meanwhile, history buffs might appreciate that we're still using a 16th century[0] term to describe a disease that is being studied 60+ years after etiological discovery[1] for which available remedies are modestly effective and not without risks [2][3].
Pretty soon, what happened to anesthesiology (one physician managing multiple extenders managing lots of technology in multiple places) is going to happen to the rest of medicine.
Physicians will adapt by carving out even narrower niches ("I only specialize in the LEFT eye..." ;). But I really don't see a they-took-our-jobs moment until people are allowed to file lawsuits against AI. Companies will never put out an AI M.D. in the wild unless there's a flesh and blood counterpart to shoulder the liability (+1 job right there).
Don't get me wrong though, I eagerly anticipate the day when going to the doctor/machine is not such a miserable soul-sucking experience... (I'm a physician.)
US > D dimer to rule in VTE, as many things can cause the latter to be elevated [1]. We almost always go straight to imaging when there's even a remote possibility of DVT.
M.D. here, one of the biggest red flags was that some patients had warfarin dosing adjusted based on these tests. As a blood thinner, warfarin requires close monitoring and accurate INR measurements to keep patients from bleeding to death.
(Plug) I shipped a dedicated OpenAI-compatible API for this, jsonmode.com a couple weeks ago and just integrated Groq (they were nice enough to bump up the rate limits) so it's crazy fast. It's a WIP but so far very comparable to JSON output from frontier models, with some bonus features (web crawling etc).