[Xmca-l] Re: FW: NYTimes: If an Algorithm Wrote This, How Would You Even Know?

[Peter Smagorinsky]

Thanks Annalisa, always a fresh perspective when you post. I have no expertise in such matters so can only speculate. These algorithms presumably are all designed by people, so we can't quite eliminate human artifactual presence from their composition. When I get an announcement from a web page such as academia.edu, I'm not expecting a person to be sending it, but rather a computer-generated, fill-in-the-blanks form notice. These statements have been around since long before computers, e.g., boilerplate documents such as wills. 

I'm mostly interested in this story as an issue of mediation. On the user end, I can mediate my conduct in relation to the computer-produced messages; I can go to academia.edu and download an interesting article, or go to a website when someone posts to Facebook and I get an automated alert. No problem there; I don't care who or what wrote the message, and would consider it cost-ineffective to have a person sitting around writing them.

The generation of a whole news story is different, and I'm still thinking about why--the greater complexity and reliability of the information are salient factors.

But from the standpoint of mediation as a function of a speaker's generating thoughts "at the point of utterance" (a phrase I take from Arthur Applebee) through the process of speaking and thus articulating inner speech into verbal text: The computer-generated text doesn't seem to have that capacity, although my dim understanding of how these things work may blind me to feedback loops that may well affect how content is generated.

Complicated stuff! People are integrating themselves with machines more and more as the technology improves, e.g., prosthetics with computers installed in them, and who knows what  in terms of google glass and other innovations. Perhaps someone out there who understands technology can help clarify some of the issues. p

-----Original Message-----
From: xmca-l-bounces@mailman.ucsd.edu [mailto:xmca-l-bounces@mailman.ucsd.edu] On Behalf Of Annalisa Aguilar
Sent: Sunday, March 08, 2015 6:44 PM
To: eXtended Mind, Culture, Activity
Subject: [Xmca-l] Re: FW: NYTimes: If an Algorithm Wrote This, How Would You Even Know?

Hi Peter,

This is an intriguing article. Thanks for posting it. I think it passed the Turing test for me, because I thought the first example was written by a person because it didn't have that much data, and the other example had too much data, but then I was comparing them, so it's not a clean Turing test.

I would like to clarify to the esteemed others on our beloved list that a feed is an automated script (which would live at XMCA server) that would ping other sites (like the way a radar works) and capture links from these offsite websites. These links (and perhaps a truncated blurb of text) would display on the xmca homepage, with the idea that there would be no need to control permissions on the server once the feed script was set up. 

At least that's the idea, it may be more complex than that (Huw??), but it is worth exploring because it means that any XMCA author could be in charge of one's own content, as a blog or any other digital content offsite from XMCA and the links would appear (ideally) at the XMCA homepage, making it a kind of clearinghouse for current events in the community in terms of other published content appearing on the web.

Of course we might want to remember that this will likely generate more discussion in the wider community, which would be a good thing, but have we the server horsepower? Just asking...

If any one is on Academia, this is happening when you get notifications that Andy has uploaded a new draft of a paper, for example (if one is following him there obviously. Andy has not become that ubiquitous just yet!!!) So instead of this feed appearing as an email (though it could do that), it appears as a list or stream at the XMCA website. 

Another thought: There is the possibility of creating filters and tags and this would enable the feed to create topical pages that list links there so that one isn't seeing a flood of unsorted links, but that they are grouped based upon their tags/filters.

Of course this isn't computer generated text as is described in the NYT article, but I realized that non-technical folks may not understand what a feed is. So I thought I'd try explaining that. 

Anyone, please feel free to add to what I've said, I don't mind.

Kind regards,

Annalisa