-
Posts
3,320 -
Joined
-
Last visited
-
Days Won
117
Content Type
Profiles
Forums
Blogs
Events
Event Comments posted by richardmurray
-
-
A COMMENT
@ProfD
On 6/3/2025 at 9:41 PM, ProfD said:Anything rogue about a computer or AI or a robot comes down to human action or inaction...culpability.
wait, I did say going rogue was possible.
On 6/3/2025 at 2:25 PM, richardmurray said:Going rogue for a computer program is when it does something it isn't designed to do absent malfunction. So when a computer program is designed to interact to humans and modulate how it interacts over time, it isn't going rogue at any moment, even if malfunction. Malfunction is malfunction , not going rogue, a computer program needs to be healed if it malfunctions. Now if a computer program is designed to play chess and chooses to interact to humans using emails. that is going rogue.
Profd you are saying, that human culpability is the sole determinant for computer program activity.
I am saying human culpability represents 99% of computer program activity, while %1 is auto induced actions of a computer program whether going rogue or other.
Delano + Troy are saying, that modern computer programs commonly called AI are 50% or more capable of auto induced actions whether going rogue or other.
One thing that most people don't comprehend is testing computer programs. In the 1950s computer programs were simple enough, you could test them completely. But by the late 1980s the most intricate software was not completely tested, and with the more intricate computer programs commonly called AI, they have not been completely tested. Even if a program is engineered to produce a random result, you can test it to check the quality of its randomness, but this is very time intensive and expensive, for modern computer programs.
But, this is why so many want to go from binary to triary. from 0 and 1 to 0 and 1 and 2. Why? Triary can check binary very fast. You can reach triary various ways, but Google has made a quantum chip, they named it Willow, which uses quantum mechanics to have more states. Quantum mechanics says, an electron's behavior has various states that can be determined by reading its position or behavior. Now in defense, this is a very expensive system. I will post about it in more detail in Black Games Elite sometime later. But a series , a large series, of willow chips could be used to completely check a computer program like chatgpt before being fed data for example. It wouldnt be able to return yes or no on every input but would be able to provide averages of the results of chatgpt for random inputs.
@Delano
well what: determines a computer program(CP) is smart or determines a CP is self directed or determines human control of a CP or determines human intentions to a CP ? These are questions I have answers to but the point is for you to comprehend your own answers.
Remember, you gave the initial premise of going rogue. correct.
At this stage the issue is, you already have a premise that computer programs, highly intricate computer programs, you call AI are not malfunctioning + designed by humans optimally [defined somewhat by you as designed to always allow a human to shut it down, designed to never reject a diagnostic subsystem or subroutine, designed to act only as the human designers intended ]. So all that is left, if said highly intricate computer programs operate other than designed, is for commonly called AI to go rogue.
I get it. The problem is you and I don't start with the same premise.
I take into account that these systems can malfunction + are designed poorly.
You said
23 hours ago, Delano said:Going rogue is not a malfunction. The computer is overstepping the restrictions the programmers have set.
I can't answer the movie scenarios however going rogue is it a malfunction. It's a problem of oversight or the computer having objectives for it's continued existence that puts human life in jeopardy.
I didn't say going rogue is a malfunction. I will sadly quote myself, which I don't like to do.
On 6/3/2025 at 2:25 PM, richardmurray said:So , going rogue is when a computer makes a choice to act that isn't within its parameters, absent malfunction/getting sick.
You suggest I don't comprehend going rogue. But in my prose I defined it similar to you.
@Troy
you said
16 hours ago, Troy said:regarding the AI conversation you tend to anthropomorphize it
citing the following I said
On 6/3/2025 at 2:25 PM, richardmurray said:can a computer program go rogue before finding its individuality.
but before I said that I said the following
On 6/3/2025 at 2:25 PM, richardmurray said:are computer programs individuals like a tree or a cat or a human?
Well, each computer program is born, age, have deficiencies with age, need checkups, or doctors. Each computer programs is an individuals. Not human, not cat, not tree, not whale, not bacteria, but computer program.
so by my own words, before the quote you used, i already stated that individuality is not anthropomorphic , as I used the word. So, your attributing to my position a false association, at least by how i read my own words.
now you say
16 hours ago, Troy said:Thinking of AI as a computer program is misleading. A program means creating instruction which that are followed.
to misleading ok. You and I already don't have a similar thinking on initial ideas, so I argue, when humans don't have the same initial ideas then the extrapolations by default will be variant, any one can call another misleading. It isn't a falsehood, but it is based on an inevitability with the variance of elemental definitions. I will restate. based on how you interpreted my words difference to how I interpreted them, i think it isn't wrong or false for you to say it is misleading, but it isn't something to prove otherwise or proselytize against, cause my elemental ideas are different than yours.
you said and define
16 hours ago, Troy said:The results knowing the inputs are 100% predictable given the same inputs.
To the definition of a program well, arithmetic programs generated unknown while design intended results in the 1900s. Probability functions, hash functions have presented unknowable while design intended results. the intricacy of modern computer programs is merely that. Intircacy, more useful, more functional, faster, more ergonomic. But the same underlying principes.
So my definition of a computer program extends wider than yours. and again, this isn't anything to discuss. I have my nomenclature, and you have yours. ok.
So we differ on how I define individuality + We differ on the definition of computer program = difference of opinion. ok. As I told Delano, his position has underlying variances to mine. That leads to different results automatically. And you can say my results are falsehoods, and I can say yours are as well. But it leads nowhere, unless we share the same elemental ideas in our arguments, which we don't.
then you say
16 hours ago, Troy said:The results can be superhuman like the protein folding or completely novel. It is not the "If this, do that" we find in classical programing.
I'd define "going rogue" as the classic "alignment problem," or worse the AI become so intelligent that it operates at such an advanced level that we would be the intellectual equivalent as mold is to humans and just as annoying.
ok... the terms "superhuman results" + "classical programming" those are terms you accept and I don't, based on our different elemental ideas to the subject. And I will add again, you can say, I am wrong for having different elemental ideas but... ok.
-
-
-
-
-
-
-
-
What My Google Dashboard Just Taught Me from alicia mccalla 06/06/2025
in RMCommunityCalendar
Posted
forum post
https://aalbc.com/tc/topic/11658-what-my-google-dashboard-just-taught-me-from-aliccia-mccalla/