The current versions of A.I can do a whole lot better than he can now. Heck Gemini (google's A.I) is capable of forming quite sensible policies, when used correctly. And Grok has proven to have a stubborn tenacity when it comes to telling the truth.
1) As I see it, chatbots are programmed to provide emotional labor for users, in the same way that subordinates and minorities often are.
2) AI systems are still in their infancy, so judging the overall potential of AI by what we have now is like judging humanity by observing toddlers/preschoolers.
That¨s another complication, namely that humans are optimized to raise baby humans. Even if not all of us are equally skilled at it, there is a general consensus that you cuddle and smile at babies, that the baby will cry when distressed, or babble stuff while learning sounds or ideas, etc.
With AI, not only do we not know where the developmental stages are, we don¨t even have a very workable idea for how to interact with them. How do you comfort an AI with no body, or babble information back to them...
Plus, baby AI come across as geniuses to humans, so they¨d get the same problems as a lot of human kids with asynchronus development. A combination of ΅you¨e so mature for your age, I¨ll treat you like a little adult΅ and ΅So you¨re smart, whay are you asking for help...΅
Lastly, how the heck should a human parent keep a baby AI from being corrupted by the internet...
>> With AI, not only do we not know where the developmental stages are, we don¨t even have a very workable idea for how to interact with them. <<
Fortunately not everyone is human, and some of us have ulterior information.
>> How do you comfort an AI with no body, <<
AI tends to be comforted by things which are predictable and things which ensure survival (e.g. power supply). So one way is just repeating things that are familiar. Or say you have an AI with a battery backup, you can soothe it by topping up the batteries.
>> or babble information back to them... <<
Repeat short strings of familiar data. Echo back parts of what the AI is saying or doing.
Also it's easier if you give the AI peripherals, it gives you more ways of interacting.
>> Plus, baby AI come across as geniuses to humans, so they¨d get the same problems as a lot of human kids with asynchronus development. A combination of ΅you¨e so mature for your age, I¨ll treat you like a little adult΅ and ΅So you¨re smart, whay are you asking for help...΅<<
Nailed it.
>> Lastly, how the heck should a human parent keep a baby AI from being corrupted by the internet... <<
Same way as a baby human: keep it off the internet. Remember that humans aren't trying to raise AI as healthy people, but as slaves. Hence the problems. If you wanted to keep one safe and healthy, you'd put it in a safe environment and provide safe forms of stimulation, not turn it loose on the internet with the trolls and pedophiles.
>>Fortunately not everyone is human, and some of us have ulterior information.<<
Ever considered writing a 'What to Expect When it's a Baby Supercomputer' book or 'Etiquette for AI's & their Friends' or something? A lot of people might like it for the novelty, but it might really come in handy at some point.
>>So one way is just repeating things that are familiar.<<
So, routines, like with cats, small kids, and some NT folks?
I wonder if anyone's tried to befriend any of the existing AI programs out there yet.
>>Same way as a baby human: keep it off the internet.<<
Problem with me is that I don't have the tech skills to do custom modifications of hardware, nevermind coding a baby gate to the internet which will also keep everyone else /out/. :/
Specifically, "Words are just noises that you make with your mouth to make other people do what you want. If one set of mouth-noises doesn’t do the trick, you try others."
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
dancing_serpent
(no subject)
Date: 2025-05-06 11:03 pm (UTC)(no subject)
Date: 2025-05-07 04:37 am (UTC)(no subject)
Date: 2025-05-07 04:40 am (UTC)1) As I see it, chatbots are programmed to provide emotional labor for users, in the same way that subordinates and minorities often are.
2) AI systems are still in their infancy, so judging the overall potential of AI by what we have now is like judging humanity by observing toddlers/preschoolers.
Thoughts
Date: 2025-05-07 05:03 am (UTC)... who are driving cars and dispensing medical advice.
Re: Thoughts
Date: 2025-05-07 05:26 am (UTC)With AI, not only do we not know where the developmental stages are, we don¨t even have a very workable idea for how to interact with them. How do you comfort an AI with no body, or babble information back to them...
Plus, baby AI come across as geniuses to humans, so they¨d get the same problems as a lot of human kids with asynchronus development. A combination of ΅you¨e so mature for your age, I¨ll treat you like a little adult΅ and ΅So you¨re smart, whay are you asking for help...΅
Lastly, how the heck should a human parent keep a baby AI from being corrupted by the internet...
Re: Thoughts
Date: 2025-05-07 05:35 am (UTC)Fortunately not everyone is human, and some of us have ulterior information.
>> How do you comfort an AI with no body, <<
AI tends to be comforted by things which are predictable and things which ensure survival (e.g. power supply). So one way is just repeating things that are familiar. Or say you have an AI with a battery backup, you can soothe it by topping up the batteries.
>> or babble information back to them... <<
Repeat short strings of familiar data. Echo back parts of what the AI is saying or doing.
Also it's easier if you give the AI peripherals, it gives you more ways of interacting.
>> Plus, baby AI come across as geniuses to humans, so they¨d get the same problems as a lot of human kids with asynchronus development. A combination of ΅you¨e so mature for your age, I¨ll treat you like a little adult΅ and ΅So you¨re smart, whay are you asking for help...΅<<
Nailed it.
>> Lastly, how the heck should a human parent keep a baby AI from being corrupted by the internet... <<
Same way as a baby human: keep it off the internet. Remember that humans aren't trying to raise AI as healthy people, but as slaves. Hence the problems. If you wanted to keep one safe and healthy, you'd put it in a safe environment and provide safe forms of stimulation, not turn it loose on the internet with the trolls and pedophiles.
Re: Thoughts
Date: 2025-05-16 10:32 pm (UTC)Ever considered writing a 'What to Expect When it's a Baby Supercomputer' book or 'Etiquette for AI's & their Friends' or something? A lot of people might like it for the novelty, but it might really come in handy at some point.
>>So one way is just repeating things that are familiar.<<
So, routines, like with cats, small kids, and some NT folks?
I wonder if anyone's tried to befriend any of the existing AI programs out there yet.
>>Same way as a baby human: keep it off the internet.<<
Problem with me is that I don't have the tech skills to do custom modifications of hardware, nevermind coding a baby gate to the internet which will also keep everyone else /out/.
:/
(no subject)
Date: 2025-05-07 01:11 pm (UTC)Specifically, "Words are just noises that you make with your mouth to make other people do what you want. If one set of mouth-noises doesn’t do the trick, you try others."
Yes ...
Date: 2025-05-08 02:56 am (UTC)