Wikipedia talk:Usage of Machine Generated Content

Wikipedia community requires guidelines on how to use and cite auto-generated content. What are the key issues? Is exact copying allowed, if the data licence permits? How to cite when such tools are used? Jsamwrites (talk) 19:41, 17 August 2018 (UTC)

• I think it depends on the context and edits in question just as for other WP:RS, but would have additional concerns. This would be a different type of source to add to the RS WPSOURCETYPES, it does not fit into SCHOLARSHIP, NEWSORG, VENDOR, BIASED, etcetera. It would be important to show if it has WP:DUE weight notable for the content shown by second-party references to it as that would help restrain any tendency for a systemic bias towards Easy/Feasible items to override any view of the Importance/Use, and also to restrain any use of Big-data for small targets. I would tend to say the accuracy would be based on the reputation of the software development organization and its equivalent to editorial control and retractions via patches, that it should be a secondary development and not self-published item, and that big data analysis is not applicable for small. I think it would help people understand if examples of what is meant by auto-generated content were stated - e.g. is it a metric or a Facebook group feed, or a web-scraping composite feed stitching together multiple items (e.g. 'trending 10' or 'top 10'), is it auto-translation, is it some automated RSS spinner rewrite, or what ? And what are the properties it has ? - is it a durable item or is it dynamically generated, perhaps interactive, possibly dynamically generated content based on cookies, is it big data analysis and if it is a big data analysis in small data , Cheers Markbassett (talk) 05:37, 29 August 2018 (UTC)
• Do not use. Auto-generated sources are not reliable sources. Often unclear where info came from, various databases are well known to have tons of mistakes, and algorithms are far from perfect. As such, do not use. Renata (talk) 14:07, 3 September 2018 (UTC)
• It depends. It should be treated like any other license-compatible content: Some of it is admissible and some of it is not. – Finnusertop (talk ⋅ contribs) 10:20, 8 September 2018 (UTC)