4
$\begingroup$

What would be the result of an AI subject to a data stream or information stream, fed to it from events in an "external world", if it could not find any pattern within that stream by the methods it has in determining patterns?

Would the response be because of the limitation of its methods in determining patterns or would you be able to conclude that the information from those events are undetermined and therefore could be stated as random?

Or would the AI be unable to ever determine using its methods whether the information fed to it was from a random undetermined world because it needed an eternity of information to determine whether the data was random or not?

So what I guess I'm asking is the pursuit of creating AI by mankind on a similar level of limitation by the accumulation of "Knowledge" over time as is the other pursuits of mankind within science etc, and is only going to be achieved simultaneously with the acquisition of "knowledge" from these other pursuits. If the determination of a simulated model comes from the physics of a determined world or not will there ever be a model because of the eternity of determining whether the information from the "external world" is determined or not.

Improve this question
$\endgroup$
1
$\begingroup$

During my master thesis I struggled with this issue.

In strict sense, a real AI would be able to say, my sensors (or some of them) are gathering data i cannot understand (yet), and would ideally have a routine what to do whith such cases and ideally one day, even understand that data. But it would most probably now, if that is happening outside in the real world, or if it's due to damaged or wrong sensors. But that broadly depends how it was programmed, so your use case would have to be further defined: what specific AI are we talking about, what can it do, what not, how is it built.

AI is actually able to reproduce
a) a subset of an adult's intelligence (speach, face recognition, etc.) but just on a case by case basis b) or, if you seak to build a full grown intelligence, it has to go thorugh the development stages as humans (eyesight, voice, imitation, learning, etc.)
So AI is still in babysteps and is not even able to reproduce let's say an adolescent behaviour, or to reproduce the human evolution of a baby to an adolescent.

So, after a long speach, to come back to your original question: At the very moment an AI is not able to decide by itself, if it has no method to analize the data, if the data analized are too random, or if one of it's sensors is giving random input. Perhaps in some years there are better methods, but at the moment, machines, as babys or childs, need adults hands to teach and show them what to do in what case and how to difference each use-case. Which often means that they would need to add new modules to the AI, which, nowadays is still far away from building anything we could really define as "intelligent" or even "artificially intelligent" (PD: Intelligence here means by ways much more than plain rationality)

I deeply recommend some books of Andy Clark concerning filosofical issues and AI: e.g. "Being there: Putting Brain, Body and World together". Andy Clark is a filosofer which takes a deeper look at AI, the human brain, cognition sciences, filosofy, psychology, sociology, etc.

Improve this answer
$\endgroup$
  • $\begingroup$ But what if your intelligence is a subset of a pure mind not withstanding my own xx $\endgroup$ – Bobs Oct 13 '18 at 20:23
  • $\begingroup$ As a sweetner that conception Is valid but open to human confusion and uncertainty a pure mind would not, but the revelation is valid but its appearence has not been observed for millenia with the excuse of a prophetic nature and the obvious deity of its prophesion. $\endgroup$ – Bobs Oct 14 '18 at 0:40
  • $\begingroup$ And i add prosperence in clarity is a measure of thought not subject to regurgitation of proposed objection truth lies not in a philossophical background but in creative venture that this statement is a product of !! Circular reasoning for what its worth captain!! $\endgroup$ – Bobs Oct 18 '18 at 18:58
  • $\begingroup$ Hi Bobs, your comments are very very difficult to unerstand. I assume you are using an automated translator. makes it difficult. All Intelligence is bound to it's input sources (where and how it obtains information) and how it can process the retrieved information. Our brains know just as much as humanity as a collective can produce (science), these are limitations of our inputs and processing power. Science has a lot of divergent directions (string theory vs quantum theory) so if an AI is programmed for string theory and something quantum mechanical happens, willl it detect it as random? $\endgroup$ – Canelo Digital Oct 22 '18 at 17:15
  • $\begingroup$ I respecfully dont believe in a concept of processing to human reciept of information as i think imagination is not a process, it could be argued that maybe imagination returns from a memory of reciept of information and a clouded perception of logic that the mind then processs and reflects an output and results in observation that could be described as random. But its too material for me and self analysis of your own mind that we only have, recursive as it is for me and everybody aswell as is language we use to communicate it. The mind as i see it is not tangeable which leads to a duality. $\endgroup$ – Bobs Nov 1 '18 at 21:39
0
$\begingroup$

I am not quite sure what you have meant by your last paragrah , but let me put it this way : what happens if an a.i fails to find a pattern in data fed to it .

since you are mentioning feeding "data" to "ai" i will narrow it down to a simple learning-model which learns from "experience" (from data).

"A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E." (definition by tom mitchell).

so if your ai model fails to find a "structure" in the data fed to it , the model is said to "underfit" provided that there is a "structure" in the data fed to it.In such a case change the model (mostly the algorithms associated with it) and then try again . This is because you are mentioning there is an infinite supply of data to it. But if there is really no "structure" in data , the data is said to be a "perfectly random". for more on that visit https://math.stackexchange.com/questions/2056780/is-it-possible-to-generate-truly-random-numbers-using-a-computer

Improve this answer
$\endgroup$
  • $\begingroup$ Does that mean that man will always be a babysitter to AI to judge the measure of performance that AI has upon an attempt of the task if the outcome of the task is known then surely the process to complete it is known and has an optimal route in which the performance is the judgement and a measure from the known route by man. A determined world would generate structure in the information it produced but having said that man perceptaully observes a random nature in events but that is judged by no method in determining that else philosophy would come to end and it's speculation would cease. $\endgroup$ – Bobs Jun 3 '18 at 0:33
  • $\begingroup$ If we're subject to a determined world how does uncertainty arise from observation and perception, perfection would be a consequence but subject to the nature of it's negation that our perception enables us and the language and structure of communication we use must be then absolute and not a product of observation , observation would then be a product of an absolute language and result in creation as a projection of that communication or I may have too much cool aid :) understanding is a limitation in communication and language which produces imagination and conception from absolution. $\endgroup$ – Bobs Jun 3 '18 at 0:50
  • $\begingroup$ Learning is not function of data or it's interpretation it's a function of reference to discontinuity and it's accumulation a fee I will always succomb to xxx $\endgroup$ – Bobs Oct 6 '18 at 0:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.