By now, you have likely juuuuuust finished the last update. Hopefully, this one will be shorter, but after these years, you know better than to think of that as guarantee…

Blastomycosis

We’ll start here, mostly because it’s new to the update. The US CDC this past weekend announced it was investigating a cluster of blastomycosis cases in employees, contractors and visitors to a specific paper mill in Michigan. The timeline is a little sketchy. I have seen some news reports claiming possible cases from February, but then others making the outbreak sound more acute. Regardless, one person has been confirmed to have died from blastomycosis, with 12 others hospitalized. A total of 21 cases have been confirmed, but at least another 75 are suspected based on symptoms.

So what is blastomycosis? Blastomycosis is typically (as it was here) a respiratory infection caused when you inhale the spores of a fungus called blastomyces. It can grow inside your lungs, causing an atypical pneumonia that for most people looks like a lingering cold/flu infection that laughs at antibiotics and antivirals. Since it’s a fungus. And those don’t touch fungi.

No, this will not turn you into a zombie, like the recent popular video game now a popular HBO series starring the actor of the moment, Pedro Pascal. If you underlying lung, heart or immune issues, yes, it can be a problem–hence the hospitalizations and death above. Reports don’t say, but you can bet the fatality had one or several of those underlying factors.

So how does one come into contact with blastomyces so as to inhale its spores?

The easiest way is to spend any amount of time in the Ohio River Valley, Great Lakes region, or southwest of the United States, where blastomyces lives, merrily eating decaying woody plant material in the soil. Yes, I realize this is MOST of our US readership.

Yes, you have all, ALL, including YOU, absolutely inhaled blastomyces spores into your lungs. It’s that prolific in these regions.

No, Chow. No. They did not die! In fact, the vast majority didn’t even notice it happened. This is for a couple of reasons. First, blastomyces would really prefer to be eating dead stuff, especially dead trees. But your lungs are warm, wet and dark, which are ideal growth conditions for fungus in general. So blastomyces, given enough head start to grow in your lungs just a little, can get confused and try to eat you from the lungs out. This is a pretty rare accident though. Usually you and blasto are passing like ships in the night. Second, your immune system, as we have mentioned before, is an evolutionarily honed whirling ball of death. Because you have to breathe (we’ll get to the evidence for that later), your lungs are a bit of a defensive liability. To mitigate that, your body defends your lungs in depth. You produce more mucus than you care to know about, even when not having a runny nose, up and down your nasal and oral passageway, all the way down towards your lungs. Air can get through this, but the sugary-protein goo that is mucus will physically trap most bacteria, viruses, and yes, fungal spores like blastomyces. Think “the Blob” movie, and that’s what it’s like for most pathogens trying to get into your lungs. Underneath the mucus river, the cells lining your lungs not only lock together in tight junctions to keep -any- physically holes plugged where nasty stuff might squeeze through

…they also have cilia, which are -organized- versions of this

…on the surface of the cell, all waving in the same direction to keep the mucus river -constantly- flowing back out. So if you wonder why you get a runny nose with a respiratory infection, that’s because your immune system is responding to infection by going overdrive with the mucus river to capture more of the stuff making you sick, and physically carry it out of you. Similarly, if you wonder why smokers get so many more respiratory infections, its because they break those cilia, and now the mucus river becomes a fetid pool of trapped and angry pathogens, because they cannot move it out any more.

Yep. This is the hard hitting content you tune in for, I know.

So even if they get past those physical barriers, any would-be respiratory pathogen, like blastomyces, must now evade your cellular immune system. If your body has seen blastomyces before, or something similar, antibodies are already floating around to it. The antibodies will bind blastomyces, physically preventing it from doing much, AND angering cells like macrophages, whose job it is to eat and digest anything covered in antibodies. Neutrophils will be spraying them with deadly oxidizing chemicals and proteins that will try to pop their cell walls and membranes. A lot of other fungi tend to anger eosinophils, but blastomyces seems to trigger neutrophils a little preferentially from what I’ve read–FWIW. Since blastomyces lives outside of cells, T-cells will mostly be co-ordinating the charge against them, particularly of antibodies. No real cause to get the varsity “Among Us” players in the T-cells revved up. Regardless, your lungs are defended in depth by an immune system that is very trigger happy, and quite homicidal to anything that is NOT you that makes it into your lungs. Hence, most blastomyces spores you inhale never get a chance to do anything.

So from that, you may have already deduced how this outbreak may have occurred in a Michigan paper mill. Especially in the last couple weeks, which have seen unusually warm weather. Yes, early reports are are pointing to the chances that wood used in the paper making process at this particular mill may have gotten wet and warm while unusually contaminated with this fungus. To blastomyces, dark, warm, wet decaying wood looks like this:

For our very confused non-US readers, this is an example of one of several chain restaurants in the United States that run as an “all you can eat” buffet. You pay one flat price for admission, and then you can make as many trips to the buffet line as you want. Does this incentivize the patrons to eat as much as possible, to get as much value out of their admission price as they can, and thus massively overdo the calories? You bet! Does this also incentivize this model of restaurant to provide cheap, high caloric foods that are easy to prepare in mass quantities so they are still able to make money doing this to stay in business? Absolutely! It’s an arms race that is possibly, maybe even probably, tangentially related to the increasing rate of obesity and type 2 diabetes.

These, not McDonalds, I would argue are the most American of American chain restaurants.

While Google tells me Old Country Buffett is apparently gone now, it still fits a little better than Ponderosa Steakhouse, so I’m taking the artistic license.

Anyways, warm, wet and woody is a buffett with an endless chocolate fondue fountain to Blasto, and they are hitting the dessert section early and often to get their money’s worth. Other famous outbreaks of other fungal disease like the previously mentioned historical examples of ergotism (St. Vitus’ Dance) have also been linked to unusually warm and wet conditions for stored rye crops prior to the outbreaks.

Blastomycosis does NOT spread person to person, and unless you happened to be working at or touring paper mills in this specific part of Michigan, your risks from blastomycosis are no more elevated than usual (depending on if you live in North America, and where you live or visit there).

Marburg

I was pausing to cover the two, count ’em two, active Marburg virus outbreaks in Africa because details have been so sparse, and honestly, they look retrospective. Sometime in February, a cluster of unexpected deaths with hemorrhagic viral symptoms was noticed in Equatorial Guinea. You may recognize Equatorial Guinea as being separate from Guinea and Guniea-Bissau, and if you want to impress your friends and family, it’s located just northwest of Gabon along the Atlantic Ocean just to the left of the Republic of Congo and Democratic Republic of the Congo as you look at the map. As near as I can tell, there were less than a dozen cases, most of them fatal, and apparently either the infections and/or deaths happened in late January. There has been no major outbreak since, but obviously the delay in announcing this is cause for concern about how actively this is being tested.

Similarly, on the opposite coast of the continent, Tanzania had its first outbreak of Marburg ever, again involving under 10 cases, most of them fatal, and as near as I can tell from reports, probably happened about a month ago. Tanzania is a direct line away from Equatorial Guinea, crossing straight through the Republic of the Congo and Democratic Republic of the Congo as you move from the Atlantic to the Indian Ocean across Africa. Yes, this is the same Congo and DRC that take the brunt of most Ebola family viruses, which includes Marburg, as you may recall. Marburg is the grand-daddy of them, and it is a little unusual for Marburg and not the slightly more common Ebola Zaire to be breaking out. However, all of these countries are very close, and all have a lot of Egyptian fruit bats, which is believed to be the reservoir of Marburg.

Like all the Ebola family hemorrhagic viruses, Marburg is highly lethal, with case fatality rates of 60-90%. Fortunately, it’s very tough to catch, typically requiring direct contact with contaminated blood and body fluids. There are a few experimental treatments, but none of the cases here were detected in time to try them so far as I am aware. There is no vaccine for the Marburg strain. There is no known epidemiologic link between these two outbreaks on the opposite coasts of Africa. There are also, apparently, not strong links among some of the cases within the countries either, suggesting there may have been some undetected community spread.

I have no idea how active, if at all, the outbreaks really are. Details are sparse, and I can’t even find much via WHO, who should probably be more on this. But you may have seen the news since our last update that CDC sent a memo to US physicians to be on the lookout for possible hemorrhagic virus symptoms from patients coming from these countries and this region, out of an abundance of caution. So I mention that here. Your chances of catching Marburg, just to get it out of the way, are equivalent to the chances you can find Zanzibar on a map.

(it’s off the coast of Tanzania, so if you were already looking at it, and are now thinking your doomed by my analogy to Marburg, no, you’re not. Your chances of catching Marburg are quite remote unless you are in roughly that “Ebola belt” of countries stretching across the middle of Africa.)

Coronavirus

<long, long, sigh>

Yeah, there’s another omicron cousin firing up. This one is XBB.1.16, which is not close enough to a droid name in “Star Wars” and is now being called “Arcturus” to make you even more Star Wars confused.

No, my Greek readership, they have not heard your quiet, long-suffering pain of having your beautiful alphabet associated with SARS-CoV-2 variants and stopped giving them Greek letter names in belated, apologetic sympathy. The WHO just gave up on naming variants period a couple variants ago.

But there are people still tracking variants, and they have been coming up with more fun, Marvel-universe/Star Wars-y names. So “Arcturus” it is. Which apparently is a Greek name, translating to “guardian of the bear”, and given to a star found near the constellation Ursa Major. So they are not done impugning you by implication, Greek readers!

The only reason this is notable is because India, where it started, is reporting that it’s a little more likely to cause conjunctivitis, or pink eye, in kids who get it. But there is a simultaneous strain of adneovirus running around over there right now too, which is a MAJOR cause of viral pink eye, so it’s not clear how much is this adeno and how much is Arcturus. Otherwise, there is NO change to the “more contagious, less severe” trend. Cases of Arcturus are trending up globally, as it is filling in for the current omicron cousins as herd immunity is starting to take its toll on them.

And that is literally all you need to know about this new variant, despite the recent headlines.

The only other COVID news of note is that the WHO revised its COVID vaccine and booster guidance. The current guidance should no surprise to you readers, emphasizing vaccination for the high risk groups. The biggest change is that the WHO said healthy young children and adolescents should be considered low priority, and suggested countries should consider cost/benefit analysis in decisions to vaccinate within this age group. Full release here.

AI Update In Brief

The AI section last time was already fairly epic in length, so I spared you the updated method that is now driving most of the major image generation algorithms. For example, NightCafe has been updated to one of this. It’s what makes DALL-E2 the “2” version from DALL-E. Google’s Imagen, still in beta, was born with this training method.

It actually makes generative adversial networks, which spawned deep fakes and less nefarious art generation AI, obsolete.

The new hotness in training AI programs that will create or modify images/video etc. according to your prompts and desires is called “stable diffusion.”

I only have a 6th grade understanding of AI, and so I will try to simplify as best I can. There are more detailed tutorials on YouTube, especially regarding the underlying math that drives this. The point is that observations and models from diffusion in the physical sciences, such as how the drop of milk diffuses through your coffee when you drop it in, were the insight behind these new models.

What is different about them is that AI algorithm is trained to take existing images of known things and add noise to them progressively. Think of this as adding “snow”, like when you tuned into a TV station back in the day, or accidentally click on the “Antenna” option when looking for the HDMI port you just plugged into on your TV. You get all that loud, noisy static flashing across your screen. The AI adds a little static to the image. Then a little more. Then a little more. Then a little more. Eventually, the final image has been reduced to nothing but static. What once was the pixel information representing a golden retriever (for some reason, this is -always- the example picture) has now been “diffused” into the noise. But the AI has learned how the golden retriever step by step shifted into noise.

Now it learns to go backwards. It starts with a prompt to draw a golden retriever and noise, or a blank piece of digital paper. It sculpts away a little noise, and compares what’s left to super noisy pictures of golden retrievers it trained on, the ones just before it all turned to noise. Then it scrapes a little more noise away. Compares again. Scrapes more noise away. Compares some more. Step by step, what’s happening on DALL-E2 after you hit the “submit button” to your prompt, is the AI is working -backwards-, sculpting from the -noise-, step wise, bit by bit, to something that resembles your prompt. Solely because it took a bunch of things similar to your prompt, broke them into noise, and now knows what noisy to less noisy to clear images of your prompt “look” like as it works backwards.

These do NOT require training two separate models to run against each other, like a generative adversarial network or GAN, and do not run off the rails trying to fool or out-guess each other for a final model like GANs sometimes do.

Now I feel like the training methods tutorial is complete. Just to warn you, though, the field is moving so fast that one of our AI scientists remarked to me last week that the “AI Methods for Dummies” session I set up with him when I first started is now completely out of date. In less than a year.

Expect that rate of change to continue. This is a fast moving space.

You may be wondering about the cost of training these models. You can find a good resource here in the Stanford AI Report, along with other discussion, examples and metrics of AI development in both industry and academia. Key quote on cost from that report:

“PaLM, one of the flagship large language models launched in 2022, had 540 billion parameters and cost an estimated $8 million USD—PaLM was around 360 times larger than GPT-2 and cost 160 times more. It’s not just PaLM: Across the board, large language and multimodal models are becoming larger and pricier.“

The increase in “parameters” or features the AI is trained to evaluate is noteworthy as well. As the parameters these larger models are getting trained against increase, there is evidence of what they are calling “emergent behavior.” These are things that the larger models can do as they are trained against more variables that the earlier, smaller models could not. For example, GPT-3 can do arthimetic. GPT-2 could not. There was a certain number of parameters and compute after which GPT-3’s ability to do basic arthimetic suddenly takes off. You can read more in an open source paper here. Before you worry too much about Skynet, again, take a look at the linked Stanford report–they have a section on how much the famous large language programs continue struggle with general reasoning measures.

Very good at certain tasks they have trained for–tend to struggle for things they were not explicitly taught to do.

I want to stress again that I do not believe the threat to humanity, or a world in which we would want to live, comes from emergence of malevolent superintelligence in AI–a “Skynet.” Where this will go off the rails happens earlier, with the predictable abuse I laid out in the last update. Humanity will pave the road to hell with its own competing “good intentions” by misusing these amazing AI tools for what they are really, really good at, but to very, very flawed human purposes.

Ben Hunt, at Epsilon Theory, has since my update come to a similar conclusion. You should read his piece too, and you can here. Not only because it agrees with my points and is therefore genius via the principal of confirmation bias–but he does it in shorter fashion and less on the mechanics of AI. All in analogy to Augustian theology. But seriously–very readable and expands on my shared concerns.

AI will be either the greatest boon to humanity of our lifetimes, accelerating our ability to discover and push fields–or it will be the greatest curse we have unleashed on ourselves. It won’t be inherent to the technology, which is merely a tool. It will be inherent to ability, or inability, to finally heed the angels of our better natures.

Socioeconomic

Totally unrelated think piece, but I came across it and portions definitely hit home. So for your musing as well.

Finally…

Your chances of catching coronavirus are equivalent to the chances that the answer to this question is “yes”:

With the caveat that this is based on expert consensus only. There are no multi-center, properly statistically powered, randomized controlled trials examining the efficacy of breathing in preventing all cause mortality or its association with aging/longevity.

You will simply have to take my word for it.

<Paladin>