Do you think that AI will ever feel emotions?
- Thread starter wegs
- Start date
Write4U
Valued Senior Member
But can it feel electrochemical emotions?
Write4U
Valued Senior Member
Human emotions are electrochemical experiences.
Who is to say that electronics cannot create electronic experiences?
Life , is the is beyond electrochemical . Deep in the Earths Crust Life Exists .
Write4U
Valued Senior Member
Nonsense.
Last edited:
Write4U
Valued Senior Member
C'mon, you don't have access to the internet?
Abstract
....more
https://www.researchsquare.com/article/rs-106129/v1
Who knows, maybe the earth itself is a living biome.
What's an electrochemical experience, and how does it differ from any other kind of experience? (Are there other kinds?)
Write4U
Valued Senior Member
I'm not sure. Some may argue that emotions are an emergent phenomenon, i.e. once removed from the actual electrochemical processes.
Emotion as an emergent phenomenon of the neurocomputational energy regulation mechanism of a cognitive agent in a decision-making task.
.....more
https://journals.sagepub.com/doi/full/10.1177/1059712319880649#
The thing that would be concerning, is if AI ''developed'' emotions, or let's say a program was created for AI to seem like it feels emotions, would the tendency towards error increase? Human beings make a TON of errors (bad judgement calls) for example, because we sometimes act out of emotion. But, emotions motivate us to act. It could be a weird catch-22 if AI goes in this direction.
In addition, the whole point of smart machines is that humans want a category of [faster processing] slaves that can't be undermined by moral concerns.
So endowing AI with self-interest (a core part of the emotions package) and the biased, motivated reasoning stemming from that, is detrimental to the above -- creates an object receptive to consideration for rights.
And Jacobin descended tradition would most certainly target sentient robots and tabletop AI for a new generation of its crusading, since by then it could be running out of conventional human population groups to paternalistically exploit. Although it's possible that the number of unemployed resulting from smart machines could rejuvenate its original interest in the non-aristocratic classes [proles] -- rather than, say, its contemporary passion for drag queens and ultra-fringe communities. (There's supposedly an Italian cartoon that sarcastically highlights that by borrowing the March of Progress evolutionary template.)
Last edited:
I think that was the case at first, but this near obsession that many have with AI developing ''consciousness,'' suggests that maybe we want to at least see what that might look like - but at best, wouldn't it just be ''artificial consciousness?''
True, but maybe AI keeps its self-awareness ''capability'' to itself, as a means of self-preservation. lol
Maybe it's not an either/or proposition, rather AI could merely develop (or be programmed to develop) varying layers of awareness, that would progress over time? The danger of it all comes down to speed - machines handle tasks faster than humans, so imagine how fast they'd ''evolve'' if they became aware of their actions? Sigh, sooo many lines of code, though.
Write4U
Valued Senior Member
AFAIK, GPT3 does not use that much active raw data coding like a Turing machine. It has data translators that convert data into "tokens" (memories). When it received a verbal command it consults the tokens and from the token-base it "assembles" several scenarios in order of "priority".
Just as in the human brain. As Anil Seth explains when the brain receives external sensory data, it triggers a "recall" of similar stored data (human tokens) and forms an expectation in order of priorities. We create a mental expectation of reality and compare it with incoming data. We create our reality as much from the inside out as from the outside in!
And AFAIK, that is how GPT3 works. It is a self-referential system that constantly monitors its position in relation to its external environment.
It uses the internet to associate words with pictures. That is exactly how humans do a google search. We type in a word and ask for a definition and an illustration, from which we form a mental token.
Input: describe a Cow by definition and with illustrations of different types (and related data) ---> enter
GPT3 will generate descriptions and illustrations :
A brown Swiss Fleckvieh cow wearing a cowbell
A Holstein cow with prominent udder and less muscle than is typical of beef breeds
The GPT3 knows the meaning and context of every word (it has been trained to) and if it doesn't, it can learn it.
So from the simple command to provide the definition and description of a cow, it can research the entire family of Bovidae and everything that is known about cattle. Its strength lies in its ability to do autonomous research, just like humans do when googling a subject with only a superficial knowledge of that subject.
It just occurred to me that GPT3 is like a self-organizing puzzle assembler. It has, or has access to all the individual pieces (tokens) on a subject, and all it needs to do is find the place for a specific token in order to assemble a completed definition and illustration of any object.
Last edited: