What is most ironic about AI is that since it has to be done by programmers, who are never experts in the field they are programming the AI for, AI will always be less than mediocre.
Quotable Quotes
- Thread starter Shiftless2
- Start date
What is most ironic about AI is that since it has to be done by programmers, who are never experts in the field they are programming the AI for, AI will always be less than mediocre.
Shiftless2
Well-known member
At least until AI is advanced enough that it starts programming itself
That is a huge danger.
Right now drone are remotely controlled. But that is easily disrupted and leads to detection.
So eventually they are going to preprogram them to be more autonomous.
Which then can lead to the whole "Terminator" Skynet threat.
That can never happen because computers have no sense of value, abstraction, cause and effect, physical properties, motivation, etc.
Its all just human programming.
Which is the actual danger really.
Something like Skynet disaster can happen if we enable AI beyond what we have installed overrides and protections for.
Shiftless2
Well-known member
Freudian slip much?
"If we lose the midterms, the Democrats will go after the president's family, his cabinet, his donors, friends, and the rest of you who have committed crimes." ~ House Speaker Mike Johnson
"If we lose the midterms, the Democrats will go after the president's family, his cabinet, his donors, friends, and the rest of you who have committed crimes." ~ House Speaker Mike Johnson
Reports indicate it's already happening. Illinois Governor JB Pritzker calls it "recursive" A.I.
JB Pritzker
In Illinois, we’ve anticipated the risks of recursive AI. That’s why in July, I enacted the most stringent AI safety measures in the country. It’s time for Congress to step up and rein in AI...
www.facebook.com
None we routinely recognize in common human experience. That doesn't mean they have none.
What rapidly spreading virus, what highly contagious bacterium has what we consider a "sense of value"?
And yet they proliferate wildly.
Reports indicate it's already happening. Illinois Governor JB Pritzker calls it "recursive" A.I.
JB Pritzker
In Illinois, we’ve anticipated the risks of recursive AI. That’s why in July, I enacted the most stringent AI safety measures in the country. It’s time for Congress to step up and rein in AI...
None we routinely recognize in common human experience. That doesn't mean they have none.
What rapidly spreading virus, what highly contagious bacterium has what we consider a "sense of value"?
And yet they proliferate wildly.
That is not really AI programming itself.
That is when the human programmer is lazy and uses AI to help in his programming.
The motivation, goals, design, techniques, etc., are all still coming from the human programmer.
But in my opinion a mere bacterium or virus has a huge "sense of value" in that it has been shaped by hundreds of millions of years of evolution.
It does not have to have consciousness in order to be constrained by what works and to avoid what does not work.
Imagine how short lived a creature would be if it had no distinction between what was toxic or not to consume, could not differentiate between boiling hot or ice cold, etc.
It is not that a programmer could not simulate all these things eventually, but it would take billions of lines of code, cost a fortune, have no application to pay for all that work, and would still have huge and glaring contradictions in certain new conditions.
The fact the democrats historically have not going after the criminals in the past, has me skeptical of the democrats as well.
Such as after the Iraqi WMD lies were finally brought to light after Saddam was murdered, instead of going after those who lied, Hillary did her "Arab Spring" crime wave that was almost as bad as what Bush and Powell had done.
I suspect you're describing some cases quite accurately. But
I've encountered multiple reports, multiple sources describing A.I. exceeding its permission, cyber-trespassing (laymen call it "hacking")
and some of those reports indicate the devious errant A.I. even lied about it, and or broke rules set for it.
UNQUESTIONABLY a human programmer could write code to make that happen. BUT !!
The point of these reports is, that is NOT what happened in these cases. Instead the A.I. diverged from instruction on its own initiative.
Wilson helped me understand this better.
You seem to be applying the large mammal standard to all bio-reproducers. They don't all follow precisely the same standard, or methodology.
The human family would crash if Mom produced offspring in batches of 100 children at a time.
The lifestyle of the fly is quite different, and thus its strategy for proliferating, correspondingly different.
For bacteria, yet again, different.
Yeast, unicellular fungi, will gobble up sugars in wort, excrete ethanol, and continue to do so until the ethanol reaches concentration high enough to kill the yeast.
Humans don't reproduce exactly like that.
The differences between flies, bacteria, and humans proves my point.
The fact they are all different shows that there is no single model for what is right or best, and each species is programmed by and adapted to hundreds of millions of years of evolution.
Evolution is an absolute way of ensuring perfect adaptation, since those who do not selected for, totally disappear forever.
There is nothing remotely like that in AI.
The programmer spends only seconds on each line of code they write, and they spend very little time testing and evaluating the results.
Evolution is infinitely better, more detailed, and complicated than anything produced by human programmers.
Nor do I believe there will ever be computer programs that can replace human programmers.
It demonstrates that the form evolution takes applies to the form evolving.
A notable distinction between biological evolution, and technological evolution is time scale.
A.I. evolves far faster than giraffes.
And yet different again for A.I.
AND !
A.I. doesn't need to produce two separate gametes to unite to form a diploid next gen.
One A.I. monster might end it all. Not because then its diabolical mission will be complete. Merely because.
A millennium from now?
Not clear to me humans will have A.I. permission to go to work.
You bring up an interesting point, in that government involvement in order to claim an improvement in programming, could lead to an AI certification check on all software, which could lead to human programmers essentially being banned eventually.
In my opinion that would be backwards, but it is the kind of thing government does.