Wikipedia:Peer review/Artificial intelligence/archive1

Artificial intelligence

 * A script has been used to generate a semi-automated review of the article for issues relating to grammar and house style; it can be found on the automated peer review page for July 2009.
 * A script has been used to generate a semi-automated review of the article for issues relating to grammar and house style; it can be found on the automated peer review page for July 2009.

This peer review discussion has been closed. feedback mainly on _content_ and weight of content

Thanks, WhatisFeelings? (talk) 20:15, 16 July 2009 (UTC)

Introductory remarks from CharlesGillingham (talk) 12:43, 18 July 2009 (UTC) Re: content and weight. This article was carefully weighted to match the content in the most widely used textbooks and histories of the field. See Talk:Artificial intelligence/Textbook survey, which records the research. The textbooks are heavily weighted towards solved problems, so there are seven or eight extra paragraphs devoted to unsolved problems. The article gives a little more weight to the social and philosophical impact of AI, but note that these sections contain the same material covered in the introductory or final chapters of the textbooks, with approximately the same weight. The same approach has been taken towards the history of AI, covered in two sections of the article (History and Approaches) and several other paragraphs. They are based major histories of AI and the short history sections in the textbooks. This scheme has been compromised by later additions of a few sections, but is still mostly intact. CharlesGillingham (talk) 09:36, 18 July 2009 (UTC) That being said, the article could use a good copy edit and one section is as yet unwritten. CharlesGillingham (talk) 12:38, 18 July 2009 (UTC)

Finetooth comments: This is interesting and is certainly broad and possibly comprehensive. Having only a layman's understanding of AI, I must limit myself to comment mainly about compliance with the Manual of Style and related guidelines. In general, the article is very well-written and contains few errors that a proofreader or copyeditor would pounce on. However, I have doubts about the "listiness" of the article and about the plethora of extremely short sections, which are causing problems with layout. Here are my comments and suggestions.


 * MOS:HEAD says in part, "Section names should preferably be unique within a page; this applies even for the names of subsections." It also says, "Section names should not explicitly refer to the subject of the article, or to higher-level headings, unless doing so is shorter or clearer." For these reasons, it would be best to avoid repeating "AI" in the heads and subheads. It will take some thought to decide on alternatives, but, for example, "History", "Philosophy", and "Research" might be OK for heads 1, 2, and 3.
 * CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)


 * WP:LAYOUT says in part, "Very short or very long sections and subsections in an article look cluttered and inhibit the flow of the prose." This article relies heavily on the use of very short sections. This makes for a very long table of contents and a cluttered appearance, and it makes it difficult to follow the guideline in MOS:IMAGES that says, "Images should be inside the section they belong to (after the heading and after any links to other articles), and not above the heading."
 * ❌ The sections are short because they are numerous. AI has a lot of subfields, and each (major) subfield deserves at least a paragraph. I'm not sure if I can break down the field of AI into just 6-10 subfields and give them a half-page each. I've taken the liberty of placing AI's subfields into three main categories (tools, problems, approaches), but none of these "topics" justify a proper sub-article; the field simply isn't structured that way. So I'll have to let this criticism stand.  CharlesGillingham (talk) 12:03, 6 August 2009 (UTC)
 * I attempted to help the situation by removing unnecessary levels in the structure of the article. CharlesGillingham (talk) 13:07, 6 August 2009 (UTC)

History of AI research
 * WP:MOS says in part, "Do not use lists if a passage reads easily using plain paragraphs." For this reason, I'd suggest turning many of the bulleted lists in the article into straight prose. The two-item list in this section could easily be converted.
 * CharlesGillingham (talk)

Learning
 * "Supervised learning includes both classification (be able to determine what category something belongs in, after seeing a number of examples of things from several categories) and regression (given a set of numerical input/output examples, discover a continuous function that would generate the outputs from the inputs)." - This might be better if broken into two or pieces. Suggestion: "Supervised learning includes classification, the ability to determine what category something belongs in after seeing examples of things from several categories. It also includes regression, the ability, given a set of numerical examples of input and output, to discover a continuous function that would generate the outputs from the inputs." Or something like that.
 * CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)

Social intelligence
 * The short list in this section could easily be rendered as straight prose.
 * CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)


 * "For good human-computer interaction, an intelligent machine also needs to display emotions — at the very least it must appear polite and sensitive to the humans it interacts with." - Here and elsewhere in the article, the spaced em dash is substituting for a semi-colon or a terminal period. Generally, it's best to use a semi-colon or a period.
 * CharlesGillingham (talk) 12:27, 6 August 2009 (UTC)

Statistical AI
 * "Russell & Norvig" - The Manual of Style deprecates the use of the ampersand unless it is part of an official name such as a corporation name. Most, if not all, of the ampersands in the article should be converted.
 * . Other ampersands (all of which appear in footnotes) are generated by the Harv family of templates. Perhaps the templates should be changed. CharlesGillingham (talk) 05:37, 6 August 2009 (UTC)

Logic
 * This section contains a list that should be easy to render as straight prose.
 * . CharlesGillingham (talk) 12:27, 6 August 2009 (UTC)

Classifiers and statistical learning methods
 * While the article is generally well-sourced, and while the sources look reliable, it's a good rule of thumb to source every paragraph as well as every direct quote, statistic, or claim that is likely to be challenged. Three of the paragraphs in this section are unsourced. Something like the "no free lunch theorem", for example, isn't common knowledge. Ditto for other unsourced paragraphs in the article.
 * Placed the unsourced material in its own paragraph for now. The rest of the material is in the main sources for the section. CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)
 * Found the material in the sources. Just had to sort out which sentence goes with which CharlesGillingham (talk) 08:58, 8 August 2009 (UTC)

Neural networks
 * The "citation needed" tag needs to be considered and addressed. Ditto for any other tags in the article.
 * CharlesGillingham (talk) 13:07, 6 August 2009 (UTC)

Specialized languages
 * The list in this section could be rendered as straight prose.
 * , by removing the list. CharlesGillingham (talk) 12:56, 6 August 2009 (UTC)

Evaluating artificial intelligence
 * MOS:BOLD says in part, "Use italics, not boldface, for emphasis in article text. Use boldface in the remainder of the article only for a few special uses... " - For this reason, I'd suggest unbolding the four listed items in this section.
 * . CharlesGillingham (talk) 12:03, 6 August 2009 (UTC)

References
 * I've never seen any reference sections in Wikipedia with citations that included bulleted lists. I see why it's tempting to do them this way, but straight prose might be better and would adhere to the Manual of Style guideline. Something like this might work for citation 103, for example: Logic: ACM 1998, ~I.2.3. Also, Russell & Norvig 2003, pp. 194-310. Also, Luger & Stubblefield 2004, pp. 35-77. Also, Nilsson 1998, chpt. 13-16."
 * ❌ I decided against this, since it produces a perfectly dense block of unreadable references. The bullet list allows readers (and editors) to easily note that each topic is covered by the most, if not all, of the central sources for the field. (As you can probably guess by my opening paragraph, this article is constantly sideswiped by people who believe that the article is biased away from or towards one or the other subfields of AI, or who come to this article from a viewpoint based in science fiction, futurism or game AI. The only "fair" solution is to provide multiple, central citations to establish the relevance of each point in the article.)  CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)


 * Page ranges take en dashes rather than hyphens; e.g., pp. 194–310.
 * There is a bot that comes by periodically and changes these. I'm not sure to what or from what, although they look fine to me. Hopefully the bot will come by again some day and fix any that are still wrong. CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)


 * I see lots of ampersands that should be "and"s.
 * As I mentioned above, these are generated by Harv, Harvnb, etc. CharlesGillingham (talk) 10:43, 6 August 2009 (UTC)

I hope these few suggestions prove helpful. If so, please consider reviewing another article, especially one from the PR backlog. That is where I found this one. Finetooth (talk) 03:45, 25 July 2009 (UTC)
 * for reviewing the article. I think I've fixed everything I can fix for now. We need to take care of the four things missing a citation. CharlesGillingham (talk) 12:27, 6 August 2009 (UTC)