omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram.
-
@ArchiteuthisFlux responding to a lighthouse is not a bad metaphor for intelligence though, because it actually involves:
(a) observation—what is this lighthouse's pattern?
(b) measurement—what bearing is it...
(c) self-knowledge—...relative to my heading?
(d) external information—where is it on the chart?
(e) correlation—conclude where i am along a line
(f) prediction—where i will be after a period of time
(g) decision-making—is that where i want to be?
(h) action—course correct as needed.@vfig Like entering the coordinates to a rake on your front lawn into the Apollo Guidance Computer then having it precisely guide you to step on it and hitting yourself in the face.
-
@ArchiteuthisFlux
"Canadian coastguard, this is the US Navy, we are on a collision course. Adjust your bearing 10 degrees south.""Nope."
"We say again, collision course, change your bearing."
"Nope."
"Canadian coastguard, we are an aircraft carrier with a support fleet of two destroyers, three cruisers, a small flotilla of support craft and a classified number of submarines. If you do not change your bearing we will take appropriate action."
"We're a lighthouse. Your call."
@_thegeoff @ArchiteuthisFlux
Several versions of this
https://youtu.be/x91iciAYcw0Someone probably did an AI slop version by now
-
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux Didn't pretty much any video game against an AI have this property where you have a goal state and you attempt to reach that goal state considering variable user input?
You know, like wind?
Why do they think this is so profound lol
We are regressing. -
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux A PID controller can do this. -
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux
They weren't serious, were they? This is just too stupid.Also I've to steal this for LinkedIn, where the main audience of people that need to see this currently is.
#ai #ai | Klaus F.
Lol, this post where someone tried to defend #AI best explains the issues we've currently with #AI. You made all of this up just to then crash into the rocks because you didn't know how a lighthouse works...
(www.linkedin.com)
-
@ArchiteuthisFlux It's like... maybe lighthouses are warning them of something.... it's actually an amazing summation of the current state of AI thinking
@morgan @ArchiteuthisFlux
We know that the AI systems serve the company that created them and so they will mindlessly direct their defenseless crew towards warnings
https://youtu.be/-6mvwOCYJjc -
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux Yeah, but if their goal is to crash into the rocks, they are nailing it.
-
It's probably just because my family's in the fishing industry that I know the secrets of the sea: for instance, boats go on the water and lighthouses go on land, and they should generally not be in the same place at the same time.
@ArchiteuthisFlux That's what I love about mastodon: these deep insights from domain experts that you can't get anywhere else.
-
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux there's that minor crashing into land issue, but also something that moves an output left/right (or up/down or hot/cold) to correct for an error is pretty much the definition of a closed-loop system.
I mean if some VC wants to give me, say, a gazzilion dollars for an op-amp or PID controller I'm all for it.
-
@ArchiteuthisFlux
"Canadian coastguard, this is the US Navy, we are on a collision course. Adjust your bearing 10 degrees south.""Nope."
"We say again, collision course, change your bearing."
"Nope."
"Canadian coastguard, we are an aircraft carrier with a support fleet of two destroyers, three cruisers, a small flotilla of support craft and a classified number of submarines. If you do not change your bearing we will take appropriate action."
"We're a lighthouse. Your call."
@_thegeoff
Hmm. The last time I heard that story it was a Spanish lighthouse. I guess it *is* a universal problem.
@ArchiteuthisFlux -
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux reading ai-bros talk about “techniques” to improve and optimise modern AI stuff, feels like living in an alternate dimension where everybody received a targeted lobotomy to remove any learnings of the past 50y on machine learning, optimisation, and information theory…
-
omg I'm reading this dumb twitter article but can't continue because I'm laughing too hard at this diagram. Like, using this as a definition of an intelligent system then crashing into the fucking rocks because you don't know wtf a lighthouse is.
@ArchiteuthisFlux the AI knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation. The guidance subsystem uses deviations to generate corrective commands to drive the AI from a position where it is to a position where it isn't, and arriving at a position where it wasn't, it now is.
-
@ArchiteuthisFlux
"Canadian coastguard, this is the US Navy, we are on a collision course. Adjust your bearing 10 degrees south.""Nope."
"We say again, collision course, change your bearing."
"Nope."
"Canadian coastguard, we are an aircraft carrier with a support fleet of two destroyers, three cruisers, a small flotilla of support craft and a classified number of submarines. If you do not change your bearing we will take appropriate action."
"We're a lighthouse. Your call."
@_thegeoff @ArchiteuthisFlux current US war department: "bomb that lighthouse"
-
@_thegeoff
Hmm. The last time I heard that story it was a Spanish lighthouse. I guess it *is* a universal problem.
@ArchiteuthisFlux@hypostase @_thegeoff @ArchiteuthisFlux I've heard it with an Irish lighthouse.
-
@_thegeoff
Hmm. The last time I heard that story it was a Spanish lighthouse. I guess it *is* a universal problem.
@ArchiteuthisFlux@hypostase @_thegeoff @ArchiteuthisFlux the story is an urban legend (check Snopes), I have seen probably more than a dozen versions (differing in the name of US ship and nationality of lighthouse crew); however, with current trend in US policies, I wouldn't be at all surprised if one day it becomes reality.
-
@hypostase @_thegeoff @ArchiteuthisFlux the story is an urban legend (check Snopes), I have seen probably more than a dozen versions (differing in the name of US ship and nationality of lighthouse crew); however, with current trend in US policies, I wouldn't be at all surprised if one day it becomes reality.
@blotosmetek @hypostase @ArchiteuthisFlux Why let the truth get in the way of a good story though
-
@ArchiteuthisFlux the AI knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation. The guidance subsystem uses deviations to generate corrective commands to drive the AI from a position where it is to a position where it isn't, and arriving at a position where it wasn't, it now is.
-
@hypostase @_thegeoff @ArchiteuthisFlux the story is an urban legend (check Snopes), I have seen probably more than a dozen versions (differing in the name of US ship and nationality of lighthouse crew); however, with current trend in US policies, I wouldn't be at all surprised if one day it becomes reality.
@blotosmetek
Oh, I'm aware, just commiserating on the underlying truth. I'd almost be surprised if there wasn't a version in semaphore and sailing ships from other Empires.
@_thegeoff @ArchiteuthisFlux -
@ArchiteuthisFlux Yeah, but if their goal is to crash into the rocks, they are nailing it.
@johnnythan @ArchiteuthisFlux They're not, that's the really sad thing. This isn't at all how neural networks in living things find their ways towards goals. You can simulate that *accurately* in about 10 lines of BASIC or FORTRAN. I've done it.
This is multi level misunderstanding: of the exercise, of the goal, AND of how to achieve it.
-
@ArchiteuthisFlux
"Canadian coastguard, this is the US Navy, we are on a collision course. Adjust your bearing 10 degrees south.""Nope."
"We say again, collision course, change your bearing."
"Nope."
"Canadian coastguard, we are an aircraft carrier with a support fleet of two destroyers, three cruisers, a small flotilla of support craft and a classified number of submarines. If you do not change your bearing we will take appropriate action."
"We're a lighthouse. Your call."
@_thegeoff @ArchiteuthisFlux as much as I want this to be true..... https://www.snopes.com/fact-check/the-obstinate-lighthouse/
