This SIDLIT 2019 roundtable discussion focused on strategies for verifying and counteracting media bias, fake news, and the various forms of half-truths that are becoming ubiquitous in our national dialog. The discussion will include examples of media bias and alternative facts as well as positive steps that educators and technologists can take to encourage critical thinking as opposed to blind acceptance of news from the plethora of sources now available.
3. Logical Fallacies
• Genetic Fallacy
• Begging the
Claim Circular
Argument
• Either/Or
• Ad hominem
• Ad populum
• Red Herring
• Straw Man
• Hasty Generalization
• Moral Equivalence
• Post hoc ergo
propter hoc
5. What’s the Solution?
“You simply must approach the
internet with skepticism.
It’s chock full of misinformation,
and a lot of it shows up in your
email inbox.” [and on Facebook]
Leo Notenboom, 18 years
programmer at Microsoft, author,
blogger and podcaster.
11. Digital Tools Journalists Use
to Fact Check
Cindy Higgins,
instructional designer
for State of Kansas,
contributor Lawrence
Journal World
Image compliments of
Changefaces.com
12. Source Vetter
• Trusted News (Google Chrome
extension) flags with icon site
you’re visiting
• Green check: Trustworthy
• Red exclamation mark: Not
trustworthy
• Also marks sites as biased,
satire, clickbait, and malicious.
13. Site categorizations
POLITICAL BIAS:
STRONGLY
BIASED
POLITICAL
LANGUAGE.
SEXISM:
DEMEANING
AND ABUSIVE
LANGUAGE
TARGETED
TOWARD A
PARTICULAR
GENDER,
USUALLY WITH
STEREOTYPES.
RACISM:
DEMEANING
AND ABUSIVE
LANGUAGE
TARGETED
TOWARD A
PARTICULAR
ETHNICITY,
USUALLY WITH
STEREOTYPES.
TOXICITY:
DEMEANING
AND ABUSIVE
LANGUAGE IN
GENERAL.
CLICKBAIT:
ARTICLE
HEADLINES ARE
DESIGNED TO
ENTICE READERS
INTO CLICKING
THE
ACCOMPANYING
LINK(S).
14. Source Vetter
• Hoaxy displays how stories from low-
credibility sources spread on Twitter.
• Often used with Botometer: app assigns
score to Twitter users based on likelihood
the account is automated to sway opinion.
• hoaxy.iuni.iu.edu
15. Site Checker
• Monitors changes online by accessing
archived repositories.
• web.archive.org
16. Site Checker
• Monitors and tracks
changes on a specific
web page.
• You select web page
section to monitor.
• Get instant alerts
when something
changes.
• distill.io
17. Image Vetter: Google Reverse
Image Search
• Right-click an image in a Google Search,
and then Search Google for Images.
• Displays similar images and websites
that contain these images.
18. Image Vetter
• Chrome Extension:
RevEye Reverse
Image Search
• To perform inverse
image search,
right-click any
image on web site.
• Download at
bit.ly/2oGv34T
20. Video Vetter
• The Washington Post: Seeing Isn’t Believing
• The Fact Checker’s Guide to Manipulated Video
• wapo.st/2K81hST
21. Video Vetter
• InVid checks reliability and accuracy of
images and video files.
• www.invid-project.eu
22. Sign up for
weekly email!
• Best Digital Tools
for Journalism by
Ren La Forme
• Supported by the
American Press
Institute and the
John S. and James
L. Knight
Foundation.
• Sign up at
bit.ly/2MltAQr
23. Examination
of Trump and
His Twitter
Feed
Ron Rohlf, assistant professor,
Informatics, Fort Hays State
University
42. Latest Concerns: Deep Fakes
Seeing is Believing (still true?)
The Camera Never Lies
(really?)
“A lie can travel half way
around the world while the
truth is putting on its
shoes.” ~ Mark Twain
43. Deep Fake
Videos
Use existing
footage of
subject merged
with computer-
generated image
of person’s face,
with dubbed
voice for audio
44.
45. Photographs are simply a crude
statement of fact addressed to
the eye. ~ Virginia Woolf
“WE BETTER LEARN HOW TO LIVE IN A WORLD WHERE
WE TRULY CANNOT BELIEVE OUR EYES.”
~VLOGBROTHERS
48. GAN Applications/Issues
Valued Uses
Imaginary Fashion
Models
Improve
Astronomical Photos
Video Game
Modeling
Reconstruct 3D
objects from Photos
Construct Voice to
Facial Image
Problematic Uses
Political Propaganda
Fake News
Fake Friend
Requests
Pornographic Uses
Blackmail
49. Deep Fake Video Examples
Obama: www.youtube.com/watch?v=cQ54GDm1eL0 (Jordan Peele
Impersonation)
Kardashian and Zuckerbery:
https://www.digitaltrends.com/social-media/kim-kardashian-
deepfake-removed-from-youtube/ (Kardashian and Zuckerberg
Deepfake videos)
More Information:
https://www.youtube.com/watch?v=Ex83dhTn0IU (Deepfake Videos
Are Getting Real and That’s a Problem | Moving Upstream)
50. Inconsistent with previous
opinions & behaviors?
Influenced by presentation
(slick vs. stumbling)?
Appeal to emotion vs.
reason?
Focus on personal attacks,
identity politics or straight
news?
What about your own
biases, preconceptions?
Possible Defenses and Considerations
51. Sources
Deep Fakes Video
www.youtube.com/watch?v=dMF2i3A9Lzw
Deep fake videos of Kardashian,
Zuckerberg and Trump
news.vice.com/en_us/article/9kxgj3/facebook-refuses-to-remove-
deepfakes-of-zuckerberg-trump-and-Kardashian
“Mona Lisa brought to life: Samsung
AI makes famous painting move and
speak” www.pocket-
lint.com/gadgets/news/samsung/148153-mona-lisa-brought-to-life-
samsung-ai-makes-famous-painting-move-and-speak
Protected by First Amendment?
uclawreview.org/2019/03/26/deep-fakes-finding-the-balance-
between-national-security-and-freedom-of-speech/
GANs www.lyrn.ai/2018/12/26/a-style-based-generator-
architecture-for-generative-adversarial-networks/
Editor's Notes
IFLScience.com recently conducted a similar experiment, publishing an article titled Marijuana Contains “Alien DNA” From Outside Of Our Solar System, NASA Confirms. The article, as of now, has over 141,000 shares, and it isn’t about marijuana or alien DNA at all – it’s an experiment to see how many shares it could attract with an outrageous headline alone. IFLScience states within the post that “We here at IFLS noticed long ago that many of our followers will happily like, share, and offer an opinion on an article – all without ever reading it.”
What is IFLScience? PRO-SCIENCE
These sources consist of legitimate science or are evidence based through the use of credible scientific sourcing. Legitimate science follows the scientific method, is unbiased and does not use emotional words. These sources also respect the consensus of experts in the given scientific field and strive to publish peer reviewed science. Some sources in this category may have a slight political bias, but adhere to scientific principles. See all Pro-Science sources.
Factual Reporting: HIGH
Notes: IFL Science is a website that publishes the lighter side of science. A word of caution with IFL Science is they do present some stories with a left of center bias. (12/12/2016) from Media Bias/Fact Check
“Hear ye my fellow citizens, to believe all prose posted on the Internet is to court folly.”
Generative adversarial network (GAN): class of machine learning using two neural networks compete with each other. The generative network generates candidates while the discriminative network evaluates them. Contest between the networks: the generative network creates an image of interest, while the discriminative network determines errors vs. true data images. The generative network's training objective is to fool the discriminative network by producing novel candidates that the discriminator thinks are not synthesized (are part of the true data distribution)
https://www.lyrn.ai/2018/12/26/a-style-based-generator-architecture-for-generative-adversarial-networks/
Generative adversarial network (GAN): class of machine learning using two neural networks compete with each other. The generative network generates candidates while the discriminative network evaluates them. Contest between the networks: the generative network creates an image of interest, while the discriminative network determines errors vs. true data images. The generative network's training objective is to fool the discriminative network by producing novel candidates that the discriminator thinks are not synthesized (are part of the true data distribution)
https://www.lyrn.ai/2018/12/26/a-style-based-generator-architecture-for-generative-adversarial-networks/