Iumenta: A generic framework for animal digital twins

(arxiv.org)
26 points sandwichsphinx | 10 comments | 19 Nov 24 03:54 UTC | HN request time: 1.645s | source | bottom
1. j-pb ◴[19 Nov 24 06:49 UTC] No.42180654[source]▶
Reminds me of: https://en.m.wikipedia.org/wiki/I_Have_No_Mouth,_and_I_Must_...

AI overloard controls your every existence until you are killed.

replies(1): >>42180958 #
2. block_dagger ◴[19 Nov 24 07:57 UTC] No.42180958[source]▶
One of the best (and incidentally one of the first) scifi stories I ever read. Certainly stuck with me.
3. thom ◴[19 Nov 24 08:45 UTC] No.42181255[source]▶
I have no idea why it needles me so much, and I will cathartically accept any downvotes, but why has the phrase “digital twin” entered our lexicon when it just means “model” or “simulation”? I’ve worked with incredibly smart AI engineers who talked about “building a digital twin” when they just meant “storing some data”. Maddening.
replies(4): >>42182320 #>>42182648 #>>42182896 #>>42182966 #
4. tupshin ◴[19 Nov 24 11:33 UTC] No.42182320[source]▶
It generally means model running in parallel with the actual system. It is not just about being able to store data, but about being able to mirror (and sometimes predict/replicate) exactly what that system is doing.

In current software parlance, this is often used in stupidly trivial ways, but digital twins have a long and important history and function

https://en.wikipedia.org/wiki/Digital_twin

replies(2): >>42182743 #>>42183794 #
5. bArray ◴[19 Nov 24 12:21 UTC] No.42182648[source]▶
I think it has been around for a while, but I agree that it seems to be pushed more now. Keeping a model in step with a real-world environment is to be fair actually quite complex, so I can see a lot of work needing to be done, and that work needing a clear name.
6. PittleyDunkin ◴[19 Nov 24 12:35 UTC] No.42182743{3}[source]▶
To use this terminology on biological systems seems more like a very lofty goal than anything realizable with current technology.

Not that there's any issue with lofty goals.

7. wink ◴[19 Nov 24 12:55 UTC] No.42182896[source]▶
In the definition (or maybe even more the usage) I know, you're missing certain properties.

E.g. you can and will sync properties to the digital twin, and back. Obviously this does not work for animals (yet?...) but the digital twin of a car can have changes that are then propagated back to the original.

I don't get any of this in any definition of "model" or "simulation" I know.

8. leonheld ◴[19 Nov 24 13:05 UTC] No.42182966[source]▶
This is my personal opinion, from working in such places: research institutes in Europe get grants very, very easily if they mention anything related to "Industrie 4.0", and one of those keywords is "Digital Twin". Another good one is "Real-Time".

Note that I'm not directing this comment specificaly at the authors of these papers (I haven't read it, just skimmed through it). Just observations from experience.

9. uoaei ◴[19 Nov 24 14:29 UTC] No.42183794{3}[source]▶
There's a huge difference between modeling only the stimulus/response, and modeling the full interior and exterior dynamics. "Digital twin" only refers to the latter.

I wholeheartedly agree that it's a safe bet that very very few digital twin projects achieve anything close to what they propose. More typically it is a flashy label to stun management. I actually worked on digital twins -- I came into ML via physics -- and once you start digging, you find the rabbit hole is deep enough that absent massive government grants or angel corporate investors you will never get anywhere close to what could be a called a digital twin.