Hey Hoffman, remember the sneezes you had in succession last winter for 2 weeks straight? I asked chatgpt and tells me is brain cancer. Are you going to start cancer therapy ASAP?
PS: for the people that still remember WebMD at the start, they would never trust a machine for full diagnosis, let alone considering this as an option
Reid Hoffman: the original LinkedIn Lunatic
“Sir, you seem to be low on vitamin C, which gave you scurvy, but Grok says that it is more likely to be an psychosomatic response to an internal conflict between the way you live your life, and the Hitler inside you waiting to be let out”
Image recognition to help radiologists find tumors is probably fine; especially since you can usually run those models locally.
These morons think ChatGPT is “conscious” and “was trained on humanity’s collective knowledge”. THAT is the problem with
AI Derangement SyndromeEDIT: aw fuck let’s not use that acronym
There are a bunch of studies that in general show there is an effect where, despite what people say and think, they inevitably start to offload decision making to AI inappropriately and it eventually makes them worse. Harvard did a study specifically around radiologists, interestingly enough.
The “only use it as an aid” seems to be a myth.
To me it seems very similar to cocaine.
AI (statistical predictive models) work best when it’s designed for a specific purpose and when the model is too challenging to derive by hand. Detecting tumors is a specific purpose, and doing so manually is challenging enough that it requires specific training. It gets a pass by me.
Predicting protein structures/drug effects: specific purpose, check. Doing it manually, yep, very challenging. Good use of AI.
LLM chatbot: purpose is unclear. Making a non-AI-based chatbot is easy and has been done before. Verdict: useless technology
Or to put it another way, use the right tool for the job don’t use the shitty multi tool that does every job passably at best. The only exception to this rule of thumb is the humble spork, but that’s a piece of engineering genius that couldn’t be replicated by AI pushers.
If you aren’t conscious its easy to think chatgpt is
That edit is a absolute 10/10 I burst out laughing
That’s deep learning, and it’s a well known and understood statistical tool.
I had a very entertaining time asking search engine AI about various bacteria when writing an open book exam.
Ask how X bacteria acts in the oral cavity, and the AI summary calls it a beneficial species
Ask how X bacteria relates to periodontal disease, and the AI summary tells you it is a pathogen of utmost importance.
It answers solely based on how you pose the question and does not even provide an accurate summary of the websites it purports to have used as sources.
Literal blatant HIPPA violation, giving personal medical info to AI companies in exchange for useless and dangerous advice
The worst part is that HIPAA has actually already allowed AI companies to do this. Epic EHR software now has built in AI chart summary support as well as AI dictation where the patient is recorded throughout the visit by the AI agent. Somone has decided that feeding patient’s PHI and voice into an AI company’s black box is somehow acceptable healthcare practice and has actively implemented it in physical healthcare facilities.
We should be able to opt out of this shit
Why? Just because it’s evil and dangerous and stupid and useless and will make shit up? Does Anyone who matters give a shit about any of that why would they?
I’m fairly certain there’s a consent specifically for use of AI at patient checkin, but I doubt the front desk staff are trained on people refusing that consent and will likely tell you that you have to sign it to be treated which I’m almost positive is not true, and may even be illegal.
An online medical service I used a while back had a consent for AI charting mixed in with the various other consent to treat paperwork, and I did have the option to decline it (and did so). During the visit, the provider once again pestered me about using AI, and I declined. They were irritated by this, because they used it to document the visit automatically, and didn’t want to have to do it themselves, but I wasn’t denied care because of it.
Do you with in healthcare? I haven’t seen an option like that in my EHR, and it does do AI summaries of patient’s charts. (To be clear, I’m not a physician, nor do I work in a clinic where I could ask a coworker about registration or “front desk” stuff, so if could be that I just don’t see it.)
At my last visit my doctor asked if I was okay with her recording for this (dunno if it was that exact company or not) and I said fuck no. I’m there to talk about sensitive shit, you wouldn’t catch me posting any of it online and you won’t catch me narrating it to a chatbot.
so literally every doctor ever (current and in the past is doing/have done malpractice because they didn’t consult AI…
who’s paying this dude to say this… ah yes, big tech, that’s who.
It is very curious rhetorical move from:
If your doctor isn’t using AI, they are incompetent and awful and should be considered malpractice
to
We shouldn’t be forcing these Big Government Regulations on the itty bitty small bean doctors who just want to help people
Techno-Libertarianism in a nutshell. It is never a serious analysis of best practices and procedures. Always some hollow appeal to legalism out of one side of the mouth and denouncement of bureaucracy out of the other. And all in pursuit of selling a new line of magic fucking beans to the rubes.
Counting the days until Dr. Oz is talking about LLMs like he talks about ginseng and acai berry juice.
deleted by creator
I would love to see an LLM doctor trained only on the TV show House.
Did you try the medicine drug?
Context: Coding in medical nomenclature refers to code blue (life threatening emergency)
AI is great at being a sort of consultant. Something to bounce ideas off of. Sometimes it comes up with something you hadn’t thought of. It’s great to confer with. However, it’s terrible to rely on. Confer with it, but don’t defer to it.
as I continue to say its hard (once using LLM’s) to stop relying off of it. after some point of using it, it becomes so insufferable that you end up reading documentation, then you rinse and repeat. I do use it for consultation and verification, but once it starts doing stuff for me, that’s when I stop using it. the tool is great until it does your job for you.
The number of competent experts who are impressed by an LLM wielded in their specified field, is as vanishingly infinitesimal as legitimate and justifiable invocations of the term ‘AI’.
Those who have expressed the greatest enthusiasm for ‘AI’ are typically the farthest removed from actual, nuanced comprehension.
It’s a grift economy built on statistically luke-warm, vibe lobotomised corpses.
deleted by creator
Yet when they get sick enough they ALWAYS go to the hospital. Same with covid. If you don’t trust doctors when you’re healthy then stop trusting them when you’re sick. Just reminds me of that “the only moral abortion is my abortion” story.
This shit is bananas
Your comment just caused some who needs FMT to receive bananas instead.
I check it out once in a while to see what’s going on, and the OpenAI people apparently seemed to try to fix the sycophancy problem by turning it into an insufferable pedant.
“I want to check… you should put pants on before leaving the house, right?”
“That’s not exactly right. Putting pants on involves putting your legs into the leg holes in the pants. After that you should zip and button any zips and buttons on your pants.”
