According to my scope of the subjects. Artificial intelligence in a very strict sense should only contain questions pertaining to how we can create truly intelligent (creative, aware, etc.) machines. Whereas data science is directly the manipulation of data in order to produce tools to make something better (image detection, intrusion detection, etc.). The separation is very blurry and I do not think that intelligence can exist without information/data, and its manipulations. However, most machine learning and deep learning demonstrated are simply conducting complex function approximations.

However, due to the popularization of machine learning and especially deep learning, the manipulation of data has created the impression of intelligent machine due to it being capable of competing with human performance in very targeted tasks (object recognition, segmentation, etc.). Evidently, the name artificial intelligence catches people's attention much more than data science or machine learning. Thus, it is common for news which is essentially generic data science/machine learning to be called artificial intelligence.

This is further demonstrated on the Artificial Intelligence site where the majority of the questions are pertaining more to data science and machine learning than truly discussing possibilities, methods or emerging work pertaining to machines capable of intelligence.

This niche can easily be encapsulated into a site that combines both Artificial Intelligence and Data Science.

Post on Data Science is found here.

| |
  • $\begingroup$ See my answer below. The Cross Validated community was apparently unhappy when Data Science launched, because of the profound overlap with an established Stack (CV.) AI is distinct because we also deal with the philosophical and social impacts, the game theory aspects of AI, and combinatorial games, which are used in AI proving. $\endgroup$ – DukeZhou Feb 21 '18 at 18:24

Though we get a lot of (off-topic) questions about data manipulation and implementation issues in general, this site was created to serve questions that aren't so quantitative. For some more info on our scope, see the help center. Admittedly, we are currently doing an incomplete job of making the scope clear to new users and handling off-topic questions. Nevertheless, it is clear from this Area 51 Discussion Zone post by a Stack Exchange community manager that this site is for AI as a science, not as a technology to be implemented:

Notice that this proposal is in the 'Science' category; not 'Technology'. Despite the creation of a Data Science site to cover this topic, the community made a sufficiently compelling case that there is a swath of questions in the academic humanities arena that are not covered by our current sites.

Social, conceptual, and philosophical aspects of AI are on-topic here, but not on Data Science, which is a more technical site. There is some overlap in architectural questions, but there is much precedent for sites' topics not being fully disjoint — Stack Overflow and Super User on PowerShell questions, for example. Combining our slightly subjective questions with Data Science's technicality would be mixing two different types of questions (and two different communities).

In short, this site and Data Science are looking at different aspects of artificial intelligence. Both sites are valuable, each with its own knowledgeable people, and it would be good to preserve the distinction.

Relevant MSE: Can Stack Exchange follow a more generic approach?

| |
  • $\begingroup$ Take a look at the new questions currently on the site. I think due to the media calling machine learning, artificial intelligence, and the unclear border between both of these topics, this site contains more off-topic questions than on-topic ones. We should just combine both sites and extend the scope to implementation and philosophical discussion. $\endgroup$ – JahKnows Feb 6 '18 at 2:44
  • $\begingroup$ Most individuals who have insight on the philosophical or ethical aspects of AI are studying the machine learning/data science. The ultimate goal of data science/machine learning is an autonomous AI. So as time progresses the data science site and the AI site will become more indistinguishable. $\endgroup$ – JahKnows Feb 6 '18 at 2:48

Hellz No! Where would people ask philosophical questions related to AI, or discuss theoretical topics? What about the Mythology of AI? (Off-topic at Stack:Mythology, but is the predominant influence re the public's perception of AI.)

Morality of AI applications is a critical issue, only increasing, as are social impacts of AI. This Stack is the forum to discuss them.

This is also the Stack for Game Theory as it relates to AI, and combinatorial games, which are inextricably related to AI, in that they are still used for AI proving.

I'd propose, as the Cross Validated community has, that Data Science should probably be rolled into that Stack, and CV should probably adopt the name "Data Science" so people know where to go for those questions. (i.e. "CV" is cool, but it's insider-ey, and noobs don't know that's the place to ask Data Science questions related to AI, and come to SE:AI.)

Do I don't think the problem is with the AI Stack at all. The humanities size of the equation, which is the core of this stack, should not be handled on a Data Science forum.

I think the solution would be to revise our "Community Guidelines" to be very clear about which questions should go to CV/DS, reposting their guidelines as a sub-section to AI's guidelines, and try to get some involvement from the CV/DS forums on which questions to migrate, so we don't accidentally migrate questions they don't want.

| |
  • $\begingroup$ This site gets a very small number of questions pertaining to the humanities regarding AI. It can easily be engulfed into a single site. Moreover, as I explained, most people with interesting insight about the ethics, mythology, etc of AI have practical experience with the subject. Thus they are well equipped to answer both flavours of questions. Why not make it more convenient by having both in a single place? $\endgroup$ – JahKnows Feb 22 '18 at 2:30

My first opinion has been "nooo!". However, skipping points as if this site must include applied AI or only strong AI, what is AI and what is data processing, ... my opinion is now "yes, just find a good name for the combined site".

The reason: this site has a low volume of questions and answers (if all off-topic was directly closed, activity will be epsilon), low number views per day, and, sorry to say that, low quality of questions and answers. Join two sites will increase the activity and the number of experts, improving all these aspects.

A good usage of tags in the new site will solve all practical aspects.

| |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .