AI Safety Memes Wiki
This is a database of memes relevant to AI existential safety, especially ones which respond to common misconceptions. Many of them were (re)posted by AI Notkilleveryoneism Memes on X. As with most memes, they are not full-fledged arguments, and regularly represent critics in uncharitable ways. Memes are fun and sometimes informative, but remember to take them with a grain of salt.
AGI
won't be powerful enough to destroy humanity
AI is “just” x (math, matrix multiplication, etc.)
Doubt on AGI capabilities (too far away, impossible, etc)
(animated version here)
Goalpost-moving (still not AGI!)
Why would AI want to be evil?
Whatever AI does is good since they’re way smarter
AI alignment
is a fringe worry
We can’t just regulate AI development
AI regulation is bad
Alignment will be solved, don’t worry
The real danger is from modern AI, not ASI
Superintelligence
An AI with cognitive abilities far greater than those of humans in a wide range of important domains.
View full definition
(myopia)
Superintelligence
An AI with cognitive abilities far greater than those of humans in a wide range of important domains.
You’re saying other risks aren’t important
Debates abt consciousness, sentience, qualia, etc.
F*ck Moloch!
Misc other takes
“AI acceleration is good”
“Yudkowskian scenarios are unlikely”
“Death is inevitable”
“Start more AI labs”
“AI is like electricity”
“I only thought abt it for 5 secs”
“AI safety
AI safety
A research field about how to prevent risks from advanced artificial intelligence.
View full definition
is a Pascal's Wager!”
AI safety
A research field about how to prevent risks from advanced artificial intelligence.
“AI safety is an infohazard!”
“You can’t predict the future exactly”
Rapid-fire bad solutions (bro just…)
Bad Yann takes
Just train multiple AIs
Just turn it off if it turns against us
*Just raise the AI like a child*