As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts.
-
@mrundkvist Wait, are you saying LLM-based AI, or SI as it's now being rebranded, is actually SLOW (Statistical Likelihood of Words)?
@tildalwave
How can I convince you that I have any mind at all when I answer?
-
@tildalwave
How can I convince you that I have any mind at all when I answer?@mrundkvist Individualize? Personalize? You know, things no LLM can do. Like, if I repeatedly ask LLM if we're there yet, it will reliably respond with the same answer no matter how many times I ask it until I spend all my tokens. If I try that with a real person, they might first respond politely, be a bit annoyed in their second reply, and smack me in the face when I lost their patience.
-
As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts. That's all it does. It's artificial but not intelligent.
You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.
@mrundkvist
> You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.Tape a knife to it and say it's going to kill us all! \s
-
As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts. That's all it does. It's artificial but not intelligent.
You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.
Spot on!
-
As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts. That's all it does. It's artificial but not intelligent.
You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.
@mrundkvist In the very early 2000s, we did have an Aibo robot dog in the house for some time. I do agree with the comparison, but I want to say that the robot dog was much cuter and did actually occasionally provide some dopamine.
-
As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts. That's all it does. It's artificial but not intelligent.
You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.
Are your feelings towards the robot any different from your feelings for a stuffed animal you had for years or towards the hamster your cats for? What about your feelings for a chat partner you communicated with for months and months? Do you believe there is a difference in the quality of those feelings?
I think we will have to answer a lot of questions in the coming years.
-
Are your feelings towards the robot any different from your feelings for a stuffed animal you had for years or towards the hamster your cats for? What about your feelings for a chat partner you communicated with for months and months? Do you believe there is a difference in the quality of those feelings?
I think we will have to answer a lot of questions in the coming years.
@SaltyMonk
Trust me, we really need to ignore our feelings for robot vacuum cleaners. -
Spot on!
-
As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts. That's all it does. It's artificial but not intelligent.
You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.
You do know that these vacuum cleaners are sleeper-agents which will be our demise in the coming robot-uprising?
SCNR.
-
As a response to your prompt, the LLM gives you the words that are statistically most likely in the order that is statistically most likely, given its training texts. That's all it does. It's artificial but not intelligent.
You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.
@mrundkvist artificial repetition, aka AR. instead of super intelligence, I say we co-opt the now defunct augmented reality acronym that died with the metaverse and we re-appropriate it to this latest hype cycle.
-
@mrundkvist
> You're reading a mind into the LLM output just the same way as you think about your robot vacuum cleaner as a house pet.Tape a knife to it and say it's going to kill us all! \s
You are right to say a major problem is human perception of AI.
I remember my first exposure to the Eliza program last millennium. Like everybody else I was bowled over by its apparent intelligence. That was until it quickly became clear the algorithm and limitation of scope. Compared to Eliza the current chat bots are literally in another world. Unless you know how they are built they will already beat the Turing test. That makes them effectively human to society in general.
If it quacks like a duck, walks like a duck than people will jump to the obvious conclusion. Soon we won'tquacks be able to prove it was an ai fake video.
Society is built on myth. This is another. Good luck in combatting that.
-
You are right to say a major problem is human perception of AI.
I remember my first exposure to the Eliza program last millennium. Like everybody else I was bowled over by its apparent intelligence. That was until it quickly became clear the algorithm and limitation of scope. Compared to Eliza the current chat bots are literally in another world. Unless you know how they are built they will already beat the Turing test. That makes them effectively human to society in general.
If it quacks like a duck, walks like a duck than people will jump to the obvious conclusion. Soon we won'tquacks be able to prove it was an ai fake video.
Society is built on myth. This is another. Good luck in combatting that.
-
J jwcph@helvede.net shared this topic
