Research Publications Newsletters Above the Fold Above the Fold - 22 August 2014

 
 

Beyond the Turing Test—The Winograd Schema Challenge

h+ magazine • 29 July 2014

Can Terry tell if it's a chatbot or just a teenage non-native English speaker? This article explains with some good examples the Winograd Schema Challenge which has been put forward as an alternative to the Turing Test. It aims to provide a more accurate measure of genuine machine intelligence. Rather than base the test on the sort of short free-form conversation suggested by the Turing Test, the Winograd Schema Challenge poses a set of multiple-choice questions that have a form where the answers are expected to be fairly obvious to a layperson, but ambiguous for a machine without human-like reasoning or intelligence.

I love this kind of stuff. If you want to brush up on the Turing Test there's the wiki article but I like this from the Internet Scrapbook which has some good archival stuff. I enjoyed reading the hullabaloo (start here) that erupted when the chatbot "Eugene Goostman" passed the 2014 Turing jury. And don't you agree that those Winograd schema questions will be a lot harder? And do you believe that's really what Eugene wants to be when he grows up? (Michalko)

Comments?

 
 

Social Good—Scale x Impact (Who knew?)

Stanford Social Innovation Review • 28 July 2014

Are you shocked that they are talking about scale AND impact? This is a short essay by two social entrepreneurs pointing out that nonprofits often focus on scale while evaluators focus on net impact. They argue we need both, and we need nonprofits and evaluators to adapt their approaches in pursuit of maximum social good.

The reason this essay jumped out at me was how relevant it seemed to many of our library discussions about shared infrastructure. The values that they attribute to funders/evaluators seem very familiar in our world—and explain for me why we seem to have so many high value and low participation projects. (Michalko)

Comments?

 
 

If correlation doesn't imply causation, then what does?

Data-Driven Intelligence • 23 January 2012

Did that stat class help? Maybe. Michael Nielsen explains that correlation doesn't imply causation. He acknowledges that this presents us with a conundrum: under what conditions, exactly, can we use experimental data to deduce a causal relationship between two or more variables? Since a lot of real world problems don't lend themselves to randomized controlled experiment there's an emerging theory of causal inference that helps establish the likelihood of this relationship.

Okay, this is old (I cannot reproduce how I came upon it recently) and it is difficult (at least after the first three sections) but it is also really informative. Read the beginning bits and you'll see the kind of paradoxes that get in the way of simple assertions of causality. I had to puzzle a bit but then it made sense. Simpson's Paradox alone is worth understanding—check out the UC Berkeley gender bias case. And xkcd. (Michalko)

 

Comments?

 
 

Fedoras, Vahs, Smugshrug

Two short ones about words and one about an annoying or endearing non-word.

Fedoras to mullets: decades of fashion words | OxfordWords

What Your Pronunciation of These Words Tells Other People | Mental Floss

The Life and Times of ¯\_(ツ)_/¯ | The Awl

Fedora came from the title of a play.

I say VACE.

I shake my head in puzzlement more than shrug. (Michalko)

Comments?

 
 

Above the Fold Quiz

According to an item in this week's News and Views section, what are elements that should be considered when constructing an outsourcing agreement for transferring born digital content from a physical medium?

Get the answer.