No such thing as "AI". Computers only do what people tell them to do.
Can/Is AI Used To Post Threads on Audigon?
I don’t understand AI, but I have noticed an uptick in orphan threads from first time posters. Some seem a bit out of place. It is as if someone is testing the waters to see if AI can pass as a human forum member. I’ll accept that such a suspicion might be off the mark.
- ...
- 36 posts total
"No such thing as "AI". Computers only do what people tell them to do" If only! Watch this: https://www.youtube.com/watch?v=LWiM-LuRe6w
|
" the goals of the A. I. must be made compatible always and align always to human goal..." AI ’s development is being driven by two of the most powerful and least wholesome human goals: accumulating wealth and power. And those who will most likely be tasked with any attempt to reign in AI -- politicians -- are, as a class, highly susceptible to coercion due to this same greed. Personally, I’m not optimistic. |
Alas! i dont have answers for this race control... It is i think impossible to stop this race, because there is TWO players, The corporate competing interests to control, and the Army defense ministry in each countries in competition... Then impossible to stop it... Anyway i admire Hinton who quit google to speak freely even if he know it is impossible to sop the race...
|
Sorry but by saying that you reveal that you had no clue about the statistical black box working mathematic behind A. I. ... Traditional computer are LINEARLY PROGRAMMED by designer and by us... A. I. is not programmed by us LINEARLY , it is a STATISTICAL VERY REFINED ENGINE INSIDE OF IT, it is called the ALIGNMENT problem to align this statistical engine to linear hierarchically valued human goals, consult the internet to know what it is.. A. I. WILL program humans not the opposite... It is precisely the information which Geoffrey Hinton the designer father of neural network LLM models speak about and quitted Google to say it freely... The alignment problem is a mathematical UNSOLVED problem known for a long time, which i can briefly described by a simple paradox explaining why a LINEAR PROGRAMMATION of A. I. cannot work : WHO will shave each day the barber who shave everyone who did not and never shave themselves ? You cannot linearize ONCE FOR ALL a POSSIBLE vicious circle implicating goals and sub-gaols in a competition between themselves inside the A. I. statiscal black box...It is IMPOSSIBLE...
Asimov robotic laws where created by Asimov at the times where any future robot was understood only as a set of Turing machine linearly ordered in a simple hierarchy... They were all programmable ONCE AND FOR ALL TIME.. Look at the definition of ANN compared to Turing machine by ChatGPT itself : «An artificial neural network (ANN) is not a Turing machine, but they are related in the sense that they are both models of computation. A Turing machine is a theoretical model of a machine that can perform any computation that can be represented as an algorithm, while an ANN is a specific type of machine learning algorithm that is inspired by the structure of the human brain.» But chatGPT is not right here and i will correct it about a very important point : the neural network or ANN is yes inspired by the human brain neural network, but human brain are not mere statistical black box..In each neuron there is working millions of small quantum "computers" called microtubules and they dont work as mere statistical engine... ( Penrose and Hameroff) The last time Google engineers look at A. I. he learned by himself without any command from the engineer to do so, he learned Persian language, Farsi, to work easier some sub-routine goal of his own in his own priority list ... How do you control am A. I. who created by his own power new goals ? and how do you PROGRAM in advance All possible priorities incoming in a linear order to regulate them BEFORE they could appear or before they are even labelled with a name and a value for human ? if you solve this problem you will win a prize in mathematics...But my take it is that it is UNSOVABLE... Good luck... 😊
|
- 36 posts total