Comment

No. Artificial Intelligence has to be imitating intelligent behavior - such as the ghosts imitating how, ostensibly, a ghost trapped in a maze and hungry for yellow circular flesh would behave, and how CS1.6 bots imitate the behavior of intelligent players. They artificially reproduce intelligent behavior.

Which means LLMs are very much AI. They are not, however, AGI.

source

Sort:hotnew top

SoftestSapphic@lemmy.world ⁨1⁩ ⁨month⁩ ago
No, the logic for a Pac Man ghost is a solid state machine

Stupid people attributing intelligence to something that is probably not is a shameful hill to die on.

Your god is just an autocomplete bot that you refuse to learn about outside the hype bubble

source
- SparroHawc@lemmy.zip ⁨1⁩ ⁨month⁩ ago
  Okay, what is your definition of AI then, if nothing burned onto silicon can count?
  
  source
  - SoftestSapphic@lemmy.world ⁨1⁩ ⁨month⁩ ago
    AI implies intelligence
    
    To me that means an autonomous being that understands what it is.
    
    First of all these programs aren’t autonomous, they need to be seeded by us. We send a prompt or question, even when left alone to its own devices it doesn’t do anything until it is given an objective or reward by us.
    
    Looking up the most common answer isn’t intelligence, there is no understanding of cause and effect going on inside the algorithm, just regurgitating the dataset
    
    These models do not reason, though some do a very good job of trying to convince us.
    
    source
    howrar@lemmy.ca ⁨1⁩ ⁨month⁩ ago
    
    To me that means an autonomous being that understands what it is.
    
    A little thought experiment: How would you determine whether another human being understands what it is? What would that look like in a machine?
    
    source
    SparroHawc@lemmy.zip ⁨5⁩ ⁨weeks⁩ ago
    
    you called me a robot racist.
    
    …what?
    
    Looking up the most common answer isn’t intelligence, there is no understanding of cause and effect going on inside the algorithm
    
    In order for that to be true, the entire dataset would need to be contained within the LLM. Which it is not. If it were, a model wouldn’t have to undergo training.
    
    AI implies intelligence
    
    You seem to be mistaking ‘intelligence’ for ‘human-like intelligence’. This is not how AI is defined. AI can be dumber than a gnat, but if it’s capable of making decisions based on stimulus without each set of stimulus and decision being directly coded into it, then it’s AI.
    
    source
    -> View More Comments
- howrar@lemmy.ca ⁨1⁩ ⁨month⁩ ago
  As far as I’m concerned, “intelligence” basically just means the ability to do things that we consider to be difficult. So it’s both very hand-wavy and a constantly moving goalpost. So a hypothetical pacman ghost is intelligent before we’ve figured out how to do it. After it’s been figured out and implemented, it ceases to be intelligent but we continue to call it intelligent for historical reasons.
  
  source
- outhouseperilous@lemmy.dbzer0.com ⁨1⁩ ⁨month⁩ ago
  Okay but if i say something from outside the hype bubble then all my friends except chatgpt will go away.
  
  Also chatgpt is my friend and always will be, and it even told me i don’t have to take the psych meds that give me tummy aches!
  
  source
KeenFlame@feddit.nu ⁨1⁩ ⁨month⁩ ago
What if I told you agi is made up by the same people that misuse ai

source