I'm not a doc or a pharmacist (though I am in med school) and I'm sure there are areas that AI could do some of a pharmacists job but on the outpatient side they do things like answering questions for patients and helping them interpret instructions that I don't think we want AI to do... or at least I really doubt an AIs ability to gauge how well someone is understanding instructions and augment how it explains something based on that assessment... on the inpatient side, I have seen pharmacists help physicians grapple with the pros and cons of certain treatments and make judgement calls about dosing that I think it would be hard to trust an AI to do because there is no "right" answer really. It's about balancing trade offs.
IDK, these are just limitations - people that really believe in AI will tell you there is basically nothing it can't do... eventually. I guess it's just a matter of how long you want to wait for eventually to come.
This paper is interesting but I want to point out there is a difference between a research paper showing that something is hypothetically feasible and something that is actually useful clinically.
Clinically, methylene blue is used to treat a different condition, methemeglobinemia and is not used to treat carbon monoxide poisoning which relies on hyperbaric oxygen therapy.
The researcher used non-human animals; it worked on them.
The hypothetical part was only that it might also work on humans.
In any case, it seems the result was good enough as a clinical trial from the point of view of veterinary medicine, in regard to those specific types of animals.
We really shouldn't be taking chemicals used on animals for veteran purposes and use them on humans too. For example, ivermectin. It's a drug meant for horses similar to another horse tranquilizer, ketamine. Can you imagine a human taking ivermectin or ketamine?!? I remember during COVID people were shooting up horse medicine and it was just really bad and upsetting, like these people were crazy. I wish Kamala had won and would have banned this horse medicines like ivermectin.
And now this whole methylene blue thing, RFK takes methylene blue. I mean, guys, this isn't even a horse veteran medicine, it's basically ink used to stain cells. I'm sorry, but everything that has an ink or pigmented color it, there is no way on earth it has a medicinal purpose. I mean. It's ink, nothing more.
> We really shouldn't be taking chemicals used on animals for veteran purposes and use them on humans too.
While I totally agree in principle, the specific substance in question (methylene blue) is used on humans already (or should I add was used, at the time of the 1933 study), and for a related emergency purpose: fixing hemoglobin that is poisoned in a certain way, giving rise to a condition called methemoglobinemia.
> And now this whole methylene blue thing, RFK takes methylene blue.
I have no idea about that; I don't follow tabloid stuff.
Methylene blue isn't a now thing; it's been known for a hundred years or more.
> there is no way on earth it has a medicinal purpose
You might be in for a surprise when you do a 15 second web search on it.
RFK playing around with methylene blue doesn't mean anything. If he happens to ascribing to it properties it doesn't have and using it for situations for which it has not been proven, he's engaging in dangerous quackery.
People kill themselves with fentanyl, yet it's an important drug, and on the World Health's Organization list of Essential Medicines: https://list.essentialmeds.org/ (scroll down to the F section).
Oh, look what else is in this list of essential medicines! Methylthioninium chloride. A.k.a. methylene blue.
Yet here you are, claiming that there is no way it has a medicinal purpose?! But you're sure you are smarter than that RFK.
> Can you imagine a human taking ivermectin or ketamine?!?
While I can only believe the ivermectin stuff because it happened (and the crossover of people who took it is pretty strong with people likely to think drinking bleach cures autism…), I 100% can believe people take ketamine because I have, and I will again - it’s fun!
Note to the curious: always do your homework. Start at Erowid and learn about any new drug, be sure to get reliably safe drugs, and the golden rule (via Rick and Morty during a Deadmau5 NYE of all places): you can always take more drugs, but you can never take less.
That's definitely a danger zone for healthy people but interestingly enough people with things like COPD may have a blood oxygen level in the 80s and while that is indicative of the disease, they may be totally stable and may not even need oxygen [1].
My grandmother's heart was completely fucked, so they'd have to adjust the alarms on the hospital monitors after checking their files when she went in. It's like "OK, well that's the problem... consults notes... Nope, apparently that is normal for her, now lets figure out what's actually wrong". It wasn't keeping regular time and it would sometimes skip, but apparently it was pumping well enough to keep her alive for several years.
Normal in humans is definitely relative and medicine has tended to assume that if we average 1000 humans (in too many cases, 1000 white college age men) that's what human normal is, which is crazy even beyond obvious problems like " people normally have 1.999 legs apparently".
Bodies are generally pretty amazing in that sense. As long as things go out of spec _slowly_, we will often adapt quite well. In the short term, we will tend to balance even fairly extreme changes out through various chemical processes and in the long term people can even develop heritable genetic changes. (E.g., how people acclimatize and have in some cases adapted to living at higher altitudes[0])
Methylene blue is actually used to treat acquired cases of a similar condition called methemeglobinemia which is when the iron in heme is oxide from Fe2+ to Fe3+ [1]. This is different from carbon monoxide poisoning which is caused by carbon monoxide binding more tightly to hemoglobin than oxygen preventing oxygen from effectively getting into your blood.
Ahh I bet that is where the confusion is. I am a physician and I have used methylene blue in severe shock and methemoglobinemia but I was a bit worried the parent comment believed it’s a valid CO treatment.
Initially suspecting that since CO poisoning and methemeglobinemia are not the same, methylene blue might not work in CO cases, I did about three seconds of web searching and found a 1933 paper about an experiment (on animals) showing methylene blue to be effective in CO poisoning.
There was quite a bit of fluff here so I only skimmed the article but the major takeaway here is that they figured out how to run protein mass-spec "by an order of magnitude" faster. This is certainly cool but I'm not sure I understand why this is on the front page of HN. What am I missing?
edit: Does anyone familiar with the field know what the significance of being able to run protein mass-spec an order of magnitude faster is? What kind of questions can we ask now that we couldn't ask before?
I used to work in proteomics and I agree. The article is written at such a high level, the actual innovation is not that clear.
Barcodes that can 9 plex samples? Thermo TMT is up to 32 plex. Multiple injections with time offset? Also not a new idea.
What is unstated is that there is no free lunch. You can go fast or you can go “deep” (high resolution/separation). Running a MS at breakneck speed does work, but you are sacrificing quantitative accuracy and the depth of proteome coverage. Which is entirely valid way run, depending on your use case. Work in a hospital and need to do 1000 samples a day? Fast and targeted makes sense. Trying to discover some novel Alzheimer’s biomarker? You want to go slowly and measure everything you can.
Ultimately this is going to be just another tool available for researchers who have to weigh the pros and cons for the particulars of their samples.
The mention of PTMs was also a bit of a distraction. Measuring PTMs is an entirely different level of sophistication for which fast MS is rarely appropriate.
Edit: I should add that Alzheimer’s was probably just used as marketing copy, but is a terrible example for ludicrous-speed-go. Human Alzheimer’s samples are incredibly hard to procure. You either need brain or spinal fluid (apparently very painful to extract from the living). For precious samples you are usually willing to operate more slowly to ensure you do a better job.
I guess a better way to put that would have been "what is the significance of being able to run protein mass-spec by an order of magnitude faster"
The article links this to Alzheimer's research but I was hoping someone on here familiar with the field would be able to point out how significant this advancement is.
Proteomics is the kind of fundamental research that can revolutionize every single area of medicine, biotech, and etc but is too central and general to have immediate applications on any specific thing.
I like the quip that AI raises the floor not the ceiling. I think it helps the bottom 20% perform more like the middle 50% but doesn't do much for people at the top.
Maybe to get an impression that they'd be performing like them - but not actually performing.
It helps me being lazy because I have a rough expectation of what the outcome should be - and I can directly spot any corner cases or other issues the AI proposed solution has, and can either prompt it to fix that, or (more often) fix those parts myself.
The bottom 20% may not have enough skill to spot that, and they'll produce superficially working code that'll then break in interesting ways. If you're in an organization that tolerates copy and pasting from stack overflow that might be good enough - otherwise the result is not only useless, but as it provides the illusion of providing complete solution you're also closing the path of training junior developers.
Pretty much all AI attributed firings were doing just that: Get rid of the juniors. That'll catch up with us in a decade or so. I shouldn't complain, though - that's probably a nice earning boost just before retirement for me.
I randomly stumbled across Tekwetu who've made a pretty good step-by-step example of coding with Claude Code, using MCPs, etc.[1]. None of the upsell or gushing. It's a pretty simple app with a backend, with a slightly complicated storage mechanism.
I was watching to learn how other devs are using Claude Code, as my first attempt I pretty quickly ran into a huge mess and was specifically looking for how to debug better with MCP.
The most striking thing is she keeps on having to stop it doing really stupid things. She slightly glosses over those points a little bit by saying things like "I roughly know what this should look like, and that's not quite right" or "I know that's the old way of installing TailwindCSS, I'll just show you how to install Context7", etc.
But in each 10 minute episodes (which have time skips while CC thinks) it happens at least twice. She has to bring her senior dev skills in, and it's only due to her skill that she can spot the problem in seconds flat.
And after watching much of it, though I skipped a few episodes at the end, I'm pretty certain I could have coded the same app quicker than she did without agentic AI, just using the old chat window AIs to bash out the React boilerplate and help me quickly scan the documentation for getting offline. The initial estimate of 18 days the AI came up with in the plan phase would only hold truye if you had to do it "properly".
It's worth a watch if you're not doing agentic coding yet. There were points I was impressed with what she got it to do. The TDD section was quite impressive in many ways, though it immediately tried to cheat and she had to tell it to do it properly.
Since then I've also provided enough glue that it can interact with the Arch Linux installer in a VM (or actual hardware, via serial port) - with sometimes hilarious results, but at least some LLMS do manage to install Arch with some guidance:
Somewhat amusingly, some LLMs have a tendency to just go on with it (even when it fails), with rare hallucinations - while other directly start lying and only pretend they logged in.
maybe, but I find that it makes it much faster to do things that _I already know how to do_, and can only slowly, ploddingly get me to places that I don't already have a strong mental model for, as I have to discover mistakes the hard way
I've only used Copilot, but this is just about exactly right. (I've only used it for Python.)
If I'm writing a series of very similar test cases, it's great for spamming them out quickly, but I still need to make sure they're actually right. This is easier to spot errors because I didn't type them out.
It's also decent for writing various bits of boilerplate for list / dict comprehensions, log messages (although they're usually half wrong, but close enough to what I was thinking), time formatting, that kind of thing. All very standard stuff that I've done a million times but I may be a little rusty on. Basically StackOverflow question fodder.
But for anything complex and domain-specific, it's more wrong than it's right.
things backed by Claude Sonnet can get a little further out than Copilot can, and when it’s in agent mode _sometimes_ it will do things like read the library source code to understand the API, or google for the docs
but the principle is the same: if the human isn’t doing theory-building, then no one is
Exactly. I'm in a situation right now where I've inherited a bunch of systems without enough documentation, and nobody knows how some things work. Sure, we've got features to build - but one of the most important things I can possibly do is make sure someone knows how stuff works, and write it down.
I think its more effective at lowering the floor. The amount of people that can't code at all but can now slap something together makes it a huge step forward. Albeit one that mostly steps on a pile of dogshit after it hits any sort of production reality.
Its like Wordpress all over again but with people even less able to code. There's going to be vast amounts of opportunities for people to get into the industry via this route but its not going to be a very nice route for many of them. Lots of people who understand software even less than c-suite holding the purse-strings.
I know someone with a PhD in biochemistry who was hired at Intel from a cancer research lab... I'm sure he sold his chemistry background well but I always thought that was an odd hire. Maybe there are just so few qualified PhDs that they'll happily take folks from adjacent fields?
Most of the senior leadership of Amazon in the early days were a bunch of randos from a formal credential standpoint. A car mechanic leading aws engineering, a musician running logistics, a chemical engineer optimizing the network etc .
Hedge funds also hire physicists and mechanical engineers
Your phrasing _drastically_ undersells the actual relevant background and experience there:
James hamilton the “mechanic” … with EE & CS degrees and time at ibm and ms.
Dave Clark the “musician” (undergrad) … and an MBA focused on logistics.
Jeff wilke the “chemist” … who worked on process optimization at honeywell and supply chains at aderesen.
So sure, might as well say DeSantis is an SDE Intern figuring out software deployments, Vosshall is an amateur aircraft EE, or marc brooker is some foreign radar engineer.
Signed, some newpaper dude who was an AWS PE doing edge networking and operations.
Chemical engineers are so good at distributed systems that it is almost a trope at this point. It is their specialty. Their entire discipline is optimizing aggregate throughput in decentralized systems with minimal coordination.
It maps 1:1 with the computer science but chemical engineering as a discipline has more robust design heuristics that don’t really have common equivalents in software even though they are equally applicable. Chemical engineering is extremely allergic to any brittleness in architecture, that’s a massive liability, whereas software tends to just accept it because “what’s the worst that could happen”.
From the tone of your post, I assume that you are a ChemE who works with CompSci folks. If what you say is true, why haven't ChemEs moved into the space and taken over? Software dev pays much better than ChemE.
Almost all of the chemical engineers I know do work in software, mostly for the money. The skillset translates to computer science relatively seamlessly. Chemical engineering is essentially computer science where you swapped atoms for bits, but far more difficult because there are only distributed systems and the background error rate is always noticeably non-zero.
I studied chemical engineering after I was already working in software, so I did it backward.
Did you study chemical engineering knowing it's applicability to software engineering?
Your observation is interesting because early ideas in object oriented design were likewise inspired by biological robustness in the face of a non-zero background error rate (see any of Alan Kay's early writings, and his Turing lecture). I wonder if half of a CS degree shouldn't also involve basic chemeng and bioeng.
Because if you need a systems designer / architect you will look for traditional credentials in the field. It’s the same reason that computer scientists cannot break into pharma despite the fact that they would really fit with the data infrastructure & processing challenges they face.
Ultimately it is all about how strict the hiring pipeline is to the credentials vs potential.
If you are able to pay off your credit card balance each month, you can earn rewards points on every purchase which is money back in your pocket. Even if it seems like a tiny amount, in the long run, you are leaving money on the table using a debit card as opposed to a credit card for most of your purchases.
The big caveat being you MUST pay the balance off each month to avoid paying interest otherwise you are losing money by using a credit card.
At least in Chile, it’s common to have 3 months of interest-free payments offered by the banks, and some stores offer up to 24 months for cards issued by certain banks.
It really helps with one-off costly purchases, like a new device or some stuff for home/garden. I never miss card payments and keep my finances organized, avoiding purchases I can’t pay off monthly.
IDK, these are just limitations - people that really believe in AI will tell you there is basically nothing it can't do... eventually. I guess it's just a matter of how long you want to wait for eventually to come.