Area 51 » Technology

The AI site didn't have enough activity during the beta, and has been closed. You can download the data dump of all questions here.

275

followers

70 Example Questions (2 closed)

active newest votes

up vote 16 down vote

added by echad Jul 22 '13 at 17:04

up vote 13 down vote

added by trav1s Jul 17 '13 at 8:39

up vote 12 down vote

added by SF. Jul 22 '13 at 11:08

not just gathering/storing one class of data indefinitely, but actually expanding scope of search basing on data received, e.g. able to turn a set of rules of a syntax into a syntactic parser, and then use said parser to parse another set of rules - written in syntax of the previous parser, to generate another, still usable one? – SF. Jul 22 '13 at 11:21

I was also thinking about this recently, but the way I framed it was, "what if the learning parameters were adapted and learned? Would this be more efficient for some classes of problems?" – trav1s Jul 23 '13 at 3:03

This is phrased similarly to a reference request as allowed on the math SE. I think reference requests are OK, but not the greatest questions. It is kind of like asking for someone else to search the literature for you. – trav1s Jul 23 '13 at 3:07

@trav1s: It is a minute of answer for an expert, and a week of search for a layman. Especially if the answer is "no, there is none" the search may take a long, long time. – SF. Jul 23 '13 at 6:38

I agree with that, but the questions are better to be from the expert than from the layman. – trav1s Jul 23 '13 at 7:47

up vote 12 down vote

added by BartoszKP Sep 18 '13 at 13:12

A very common question on SO site, with many different answers and scenarios. Probably such questions should belong here. – BartoszKP Oct 16 '13 at 12:43

up vote 11 down vote

added by Othman A.A. Jul 24 '13 at 13:08

up vote 11 down vote

added by trav1s Aug 1 '13 at 3:02

up vote 11 down vote

added by Quonux Aug 1 '13 at 12:52

up vote 11 down vote

added by nijansen Aug 6 '13 at 11:16

up vote 11 down vote

added by Maecky Aug 25 '13 at 16:10

up vote 11 down vote

added by Luiz Vieira Oct 26 '13 at 23:04

up vote 10 down vote

added by trav1s Jul 17 '13 at 3:07

up vote 10 down vote

added by e-sushi Jul 26 '13 at 11:14

up vote 10 down vote

added by echad Jul 26 '13 at 11:36

up vote 10 down vote

added by Quonux Jul 28 '13 at 0:13

"better" is ambiguous. This question lacks precision, however the topic is interesting. Maybe it would be helpful to choose a particular criterion. – BartoszKP Aug 17 '13 at 21:45

up vote 10 down vote

added by e-sushi Jul 28 '13 at 11:10

up vote 10 down vote

added by trav1s Aug 1 '13 at 2:43

up vote 10 down vote

added by trav1s Aug 1 '13 at 2:57

up vote 10 down vote

added by Keshav Srinivasan Aug 8 '13 at 15:16

Good question for philosophers and science fiction fans, but in general completely non-answerable. What answer do you expect? "2 years", "really really close", "we're very far", "10 kilometers"? In fact this asks to predict the future, and also to write almost a book chapter defining the precise version of Turing test one has in mind and what exactly does "passing" mean. – BartoszKP Nov 6 '13 at 23:03

@BartoszKP: Quite answerable if we're to create a metric based on a standarized test and extrapolate. Test many machines against a statistically significant number of people. Each machine will have a different success ("recognized as human") rate. Distance between current best rate and 50% (machine passes for human better than real human) is one simple, objective "how close we are". Create a graph of machine creation date vs success rate, find trend curve. Extrapolate the curve, find point where success rate exceeds 50% and you have a less reliable but easier to grasp time estimate. – SF. Dec 2 '13 at 9:16

up vote 10 down vote

added by echad Aug 15 '13 at 14:49

up vote 10 down vote

added by Tim Winter Aug 15 '13 at 16:42

up vote 10 down vote

added by JDong Aug 24 '13 at 7:26

up vote 10 down vote

added by nijansen Aug 25 '13 at 14:47

up vote 10 down vote

added by Maecky Aug 25 '13 at 15:46

up vote 10 down vote

added by Nikolaus Aug 25 '13 at 17:56

up vote 10 down vote

added by Nikolaus Aug 25 '13 at 18:05

up vote 10 down vote

added by lejlot Aug 28 '13 at 15:26

up vote 10 down vote

added by danijar Sep 3 '13 at 20:44

This is pretty similar to this other question area51.stackexchange.com/proposals/57719/… – trav1s Oct 29 '13 at 7:12

@trav1s I'd argue that this is a better one, because it's more precise. "learning new ways of learning" seems completely undefined to me. – BartoszKP Nov 2 '13 at 16:03

up vote 10 down vote

added by Jarno Sep 9 '13 at 18:54

This kind of question is looking more for a list than a solution to a specific problem. – Adrian Collado Oct 3 '13 at 13:50

What is a "good" language anyway? Is there any language that is not a good language to program an AI in? – Simon André Forsberg Oct 29 '13 at 0:30

up vote 10 down vote

added by Jarno Sep 9 '13 at 19:01

closed as not a real question by Robert Cartaino Nov 4 '13 at 13:37

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form.

All? Do you plan to become a wikipedia/database? There are at least few hundreads of answers for such question. – lejlot Oct 27 '13 at 10:03

This is fundamentally a really broad question and creating this type of all-encompassing list by polling the community for every possible contribution isn't really what we do. See cstheory.stackexchange.com/about – Robert Cartaino Nov 4 '13 at 13:37

up vote 10 down vote

added by Jarno Sep 9 '13 at 19:04

Wording seems rather risky here - this kind of question suggest the answer, which in many cases is simply wrong. Better question would be "What are the pro's and con's ..." then current "Does .... have any pro's" – lejlot Oct 27 '13 at 10:02

up vote 10 down vote

added by echad Sep 20 '13 at 14:06

[+1] Nice question... will be interesting to read the answers to that. ;) – e-sushi Oct 3 '13 at 22:55

up vote 10 down vote

added by Adrian Collado Oct 3 '13 at 13:57

I assume that such question would be specified to one particular problem in its full form? Such questions do not have one, generic answers, but rather the ones starting with "it depends on many factors" :) – lejlot Oct 27 '13 at 10:06

up vote 10 down vote

added by JDong Oct 3 '13 at 16:45

up vote 10 down vote

added by BartoszKP Oct 6 '13 at 9:41

up vote 10 down vote

added by BartoszKP Oct 8 '13 at 12:54

up vote 10 down vote

added by Cam Jackson Oct 13 '13 at 10:03

up vote 10 down vote

added by kdzia Oct 30 '13 at 15:54

Good question! Very often someone just uses AI approaches to a problem, just because it's "cool" or something, ignoring the fact that a problem is perfectly tractable by classical, direct methods. – BartoszKP Nov 6 '13 at 10:20

up vote 10 down vote

added by kdzia Oct 31 '13 at 11:24

Cool recurrence :D – BartoszKP Nov 7 '13 at 11:33

up vote 10 down vote

added by Meterbuchus Nov 9 '13 at 19:36

How could a mathematically competent person do it? Mimicking a human personality doesn't sound like a solved problem (unless I misunderstood) – David Robinson Nov 17 '13 at 22:24

Apple's Siri but with a more specific design – Meterbuchus Nov 18 '13 at 20:13

"easiest" - vague ; "mathematically challenged" - vague ; "AI program/algorithm" - vague ; "mimic" - vague ; "human personality" - vague. While it's not a inherently very bad question it seems poorly worded, and has so many possible interpretations that it's extremely broad. – BartoszKP Nov 25 '13 at 11:54

Mathematically Challenged - A lot of people that have a fascination with Artificial Intelligence are not necessarily fluent with the mathematical concepts such as Calculus but want to get involved. – Meterbuchus Nov 27 '13 at 19:17

Yes, I can see them here posting questions, instead of trying the Science Fiction site. Sorry, for being rude. – BartoszKP Nov 29 '13 at 17:35