Just wanted to say I don't have a really concrete explanation for this, but for an environment of 10 variables, the metric for overall variance should exercise those 10 variables as much as possible too. Rules that can be determined (if they exist) between variables x,y,z etc are the steps to achieve higher variance, ie if the rules aren't known between variables, then it is unlikely to make progress in more complex relationships between variables without meeting the more basic rules between variables.
I'm open to any ideas on a good definition for this, especially one that could be coded ie for simple simulations of input variables.
Here are some comparison states going from low variance to higher variance:
example1:
a.stationary object b.object moving with fixed velocity c.object accelerating with fixed acceleration d.object with changing acceleration (jerk)
In example1:
b has more overall variance than a, since there is a new variable that \ is changing.
c has more overall variance than b, as the velocity itself is changing
d has more overall variance than c, as the acceleration is changing
example2:
a.quadcopter hovering b.quadcopter flying in a straight line c.quadcopter flying in a straight line, while identifying objects in the flight path d.quadcopter negotiating a varying course to avoid objects in the flight path
In example2:
b has more overall variance than a as the position variable as well as velocity variable are changing
c has more overall variance than b, as there is a more complicated set of variables taken into account
d has move overall variance than c, as the required manovers to avoid obstacles create more variance in the control system as well as position/velocity/acceleration etc
These examples imply that for maximum overall variance, systems should exist with dynamic stability, so that as many variables as possible are actively changing, yet at the same time in a way that is stable, as if it is unstable, that implies that the system will not maintain its dynamic behaviour, and thus lose overall variance.
This is similar to life already, ie metastable states.
If you want a bot to not do stupidly dangerous things, you have to limit its actions, ie reduce its variance. If Jamie can't figure out that lethal actions are a variance too far... but then he can't even work out what NI is when compared to AI.
If Jamie can't figure out that lethal actions are a variance too far... but then he can't even work out what NI is when compared to AI.
Hi,
Self limited variance is the only imposed limit that can give intelligence, artificial or otherwise. Externally setting limits is destined to just be hacking with bots as you say.
The contrast between Natural Intelligence and Artificial Intelligence does seem to have escaped Jamie, who does seem have been short-changed of any kind of intelligence.
If he did have a brain in his head he'd realisie that in NT's sentence "So basic logic isn't your thing either. You need to master NI before trying AI." his expansion doesn't work.
"Not Intelligent" is not something anybody could master.Jamie is a master of not being intelligent, but that's a another kind of property.
I wonder why Jamie bothered to post that. He's got no clear idea of what "v ariance" or "intelligence" mean, so he's just playing with words in almost the same way that infants make mud pies, with the difference that infants d evelop their manual skills by making mud pies, while Jamie drains his very limited stock of credibility by posting propositions that don't mean anythi ng.
"variance" or "intelligence" mean, so he's just playing with words in almos t the same way that infants make mud pies, with the difference that infants develop their manual skills by making mud pies, while Jamie drains his ver y limited stock of credibility by posting propositions that don't mean anyt hing.
t "variance" or "intelligence" mean, so he's just playing with words in alm ost the same way that infants make mud pies, with the difference that infan ts develop their manual skills by making mud pies, while Jamie drains his v ery limited stock of credibility by posting propositions that don't mean an ything.
"Neurological" disorder. Jamie is too dumb to use a spell-checker.
That's possible. Many lunatics have quite accurate perceptions of most of the world. Dimbos like you don't know enough about the world to have an accurate perceptions of any of it.
Jamie is easily amused by simple exercises. He can't follow anything particularly complicated, which leaves him feeling left out and resentful a lot of the time.
Jamie might be able to observe natural intelligence in action, but he lacks the wit to comprehend what it does, or to recognise its presence or absence.
Jamie does like to be rude about people who are cleverer than he is. He doesn't have to search hard to find them.
t "variance" or "intelligence" mean, so he's just playing with words in alm ost the same way that infants make mud pies, with the difference that infan ts develop their manual skills by making mud pies, while Jamie drains his v ery limited stock of credibility by posting propositions that don't mean an ything.
This time Jamie managed to spell neurologically correctly. Sadly for him, m y fixing my software won't make him any brighter, though it might save me f rom wasting time reminding the group - again - what a half-wit he is.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.