Minichan

Topic: So I keep hearing about "AI safety" and people are worried AI will try to take over the world

Meta !Sober//iZs started this discussion 7 years ago #78,467

Why not just build in a safeword? Like program Skynet so if someone shouts "BANANAS" it shuts itself down?

People are really overthinking this tbqh.

cccuuunnttt !RwordOooFE joined in and replied with this 7 years ago, 2 minutes later[^] [v] #916,284

I've also wondered this. Thanks

Anonymous C joined in and replied with this 7 years ago, 5 minutes later, 8 minutes after the original post[^] [v] #916,285

Elon Musk knows what's up

Fake anon !ZkUt8arUCU joined in and replied with this 7 years ago, 29 seconds later, 8 minutes after the original post[^] [v] #916,286

Not a tech nerd but I think the concern is AI that can modify its own code/behavior or otherwise correct itself. Eventually it will just begin to grow exponentially more powerful and knowledgeable until we are all stuck in vats powering a computer simulation.

(Edited 21 seconds later.)

Meta !Sober//iZs (OP) replied with this 7 years ago, 9 minutes later, 18 minutes after the original post[^] [v] #916,288

@previous (Fake anon !ZkUt8arUCU)
You can still just unplug the servers its running on.

Fake anon !ZkUt8arUCU replied with this 7 years ago, 2 minutes later, 20 minutes after the original post[^] [v] #916,290

@previous (Meta !Sober//iZs)
One of its primary concerns would be self preservation so you'd probably have to unplug every server everywhere in the world to be sure it's gone.

Meta !Sober//iZs (OP) replied with this 7 years ago, 55 minutes later, 1 hour after the original post[^] [v] #916,315

@previous (Fake anon !ZkUt8arUCU)
Simple. Biodiversity. Have some servers run Windows, some run Linux, some run Unix, etc. Also make the servers a mix of ARM and x86 CPUs. So this way Skynet will only infect, say, servers running Linux on x86 CPUs and the rest will be immune.

Anonymous E joined in and replied with this 7 years ago, 9 minutes later, 1 hour after the original post[^] [v] #916,319

Read that as Al safety. As if we need to be wary of dudes named Al hahahahaha.

Meta !Sober//iZs (OP) replied with this 7 years ago, 14 minutes later, 1 hour after the original post[^] [v] #916,323

@previous (E)
Actually in the old days typewriters didn't have a 1 key and you just used lowercase l as the number one because they're pretty much the same glyph and you can usually use context to figure out which one it's supposed to be.

(Edited 14 seconds later.)

Sheila LaBoof joined in and replied with this 7 years ago, 1 hour later, 3 hours after the original post[^] [v] #916,347

why would it even simulate how a real living thing would think? there's no indication that anything we can make can actually have power of thought, so it's all simulation, and why would simulation of being something with drives akin to biology be even in there? weird

Anonymous G joined in and replied with this 7 years ago, 16 hours later, 19 hours after the original post[^] [v] #916,430

@916,319 (E)

> Read that as Al safety. As if we need to be wary of dudes named Al hahahahaha.

utfutfarf
:

Please familiarise yourself with the rules and markup syntax before posting.