Originally published at: https://boingboing.net/2020/11/23/scientists-are-using-to-ai-to-recreate-the-stench-of-16th-century-europe.html
…
Cool. Now do the Great Stink!
It’s… interesting? I guess? But maybe $3 million could be used in better ways than recreating old stenches? Hey, I love anthropology and archeology, but this feels like a really silly way to spend money to me.
€3 million isn’t that much money in the grand scheme of things and once we go down that route we are very quickly on the slope of asking why we should fund any scientific endeavour without any immediately apparent practical use.
I’ll venture a couple guesses: gangrenous rot and feces will both play major roles in the eventual aromatic profile they develop.
I am pretty sure it would have been hard to get funding for Boolean Algebra.
It’s an interesting exercise, but since bacteria were and are one of the major smell producers, and bacteria evolve quickly and have had 400 years to change the smells they produce since then, what the past smelled like is fundamentally unknowable. There are also all the smells that never get mentioned because they are so common that everyone knows them. The cooking smells, the animal smells, the smells of leathers and fabrics will have been different, and the unknown smells, of course, will remain unknown.
Often, when two things combine the smell of the new combination is totally different from either of the original two. Little kids discover that when they mix stuff together during play, and perfumers make a lot of money out of some of those new scents. And it’s another sources of unknown past smells that will never be recovered.
One of the most striking things about the third-world countries I visited 50 years ago was the rich variety of unusual odors. By no means were they all unpleasant. Also saw a lot of physical deformities.
Probably a rhetorical question, but why would you want to do that? Yuck!
Nothing good can come from this.
Training a computer to recreate some of the smells from one of the smelliest, least hygienic times in human history, when they threw sewage out the window onto the street, and bodies from the plague decomposed in the open roads?
We need a smell equivalent of the term nightmare fuel.
Ren Fair authenticity.
Actually reminds me of why smell will never make it into any kind of VR entertainment. Action movies would become significantly less popular if the noise levels were accurate and you could also smell the smells and in the case of first person, feel the pain. And of course, any historical drama where indoor bathrooms are a thing, but plumbing is not.
I once met someone who’s job description was ‘a nose’.
She worked in the manufacture of perfumes and had some restrictions on her daily life. Not smoking was one, but I think she had some restrictions around food and alcohol too.
So even if you personally don’t regard smells as particularly interesting or important, to some people they are very important.
I guess that’s true of most things in life.
Afterthought: Given the amount of airtime perfume brands buy in the run up to Christmas, that industry must be worth bazillions.
Don’t you have to have a library of real things to train AIs on? In this case, wouldn’t that require an input artificial nose and output smell-generator of some sort?
Finally, a positive use case for AI!
As I was taking a walk in the neighborhood the other evening, it occurred to me that the smell of wood smoke was probably pervasive in the past. Given how much smell is generated by a few suburban chimneys, I can only imagine what it was like when pretty much everything was based on wood fires: heating, cooking, steam engines, etc. Everyone’s clothes and hair must have smelled like smoke all the time, kinda like when you go on a camping trip. Maybe that would have deadened your olfactory senses to all the poo and BO and stuff.
(Yes, I’m aware that there were places that burned coal primarily, which must have been even worse. I’m thinking mainly about the West Coast where I live where coal wasn’t a thing.)
I think the AI part here is simply machine learning for accurate NLP to find the descriptions of smells in digitised source texts.
Ab.so.lutely.
This is from a text conversation I had with a group of friends a few days ago where we talked about this very article. All of them are scholars of the early middle ages in some fashion.
Though if experimental archaeology has taught me anything it’s that everyone and everything smelled of wood smoke. And tar if you’re anywhere near a ship.
I’m only copying my part because, you know, privacy. But we agreed that it would have been that and a barnyard smell.