Number of Posts: 10
Posts 1 - 10
Teaching A.I. Systems to Behave Themselves
Newspaper | The New York Times
Date | 13.8.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, Facebook, Google, threat
Summary | Artificial intelligence systems have made huge development leaps in recent times but there is still a lot of learning to do. The image recognition AI assistants of Facebook and Google demonstrate how, on the one hand, they can recognize a lot of images correctly if they have had enough data to learn from and, on the other hand, how it still makes bold mistakes. It suffices to manipulate a few pixels and the AI system gets confused. Developing AI systems not only takes a lot of data but also trial and error phases which are monitored and taught by human teachers.
Image Description | Programmers working on computers.
Image Tags | computer/laptop, female(s), male(s)
Germany Tells Sites to Delete Hate or Pay Up
Newspaper | The New York Times
Date | 30.6.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, censorship, cyberbullying, Facebook, fake news, law, threat
Summary | Germany has the strictest policies when it comes to illegalizing slanderous, threatening, and extremist language from public spaces. Germany has just passed a law that allows them to fine Facebook as much as 57 million dollars if they do not remove offensive content quickly enough from the platform. While some may say this is censorship , German lawmakers claim that respectful online encounters are a necessity for free speech to thrive. Facebook is now working on improving the flagging process for offensive material and are also using artificial intelligence to remove fake news.
Image Description | Blurry man looking at a smartphone with the Facebook logo in the background.
Image Tags | Facebook, logo, male(s), smartphone
YouTube Sets Policies To Restrict Extremism
Newspaper | The New York Times
Date | 18.6.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, Google, marketing, threat, YouTube
Summary | Google has been using artificial intelligence to weed out offensive videos from YouTube and take them down. It is quite good at detecting nudity, graphic violence, and copyright violations. However, other less straightforward offensive material remains on the platform such as cultish sermons by extremist muslims. These are however not being monetized by displaying advertising next to them.
Image Description | An image of the London Tower and a portrait of a man.
Image Tags | female(s), male(s)
Facebook Will Use Artificial Intelligence to Uncover Extremist Posts
Newspaper | The New York Times
Date | 15.6.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, censorship, Facebook, politics, threat
Summary | Facebook has been urged by both users and politicians to do more to combat extremist content on their platform. It is Facebook's responsibility to monitor the content they allow so as not to provide a safe space for extremists. Facebook has announced that they plan to employ artificial intelligence to help them flag extremist content.
Image Description | An image of a man and blurry silhouettes standing under a Facebook logo.
Image Tags | Facebook, female(s), logo, male(s)
Father in Thailand Kills 11-Month-Old Daughter Live on Facebook
Newspaper | The New York Times
Date | 25.4.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, censorship, Facebook, law, threat
Summary | Since Facebook has enabled its livestream functions many crimes have been broadcasted on Facebook. Facebook still struggles to take such offensive content down. Recently a man livestreamed himself killing his daughter and the recording stayed online for twenty hours. Facebook needs to improve its artificial intelligence mechanism to flag such content faster.
Image Description | Reuters image of people crying.
Image Tags | female(s), male(s)
Daily Report: Facebook Struggles With Being Responsible
Newspaper | The New York Times
Date | 18.4.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, Facebook, law, pornography, threat
Summary | Facebook and other social media have a huge problem with inappropriate or criminal content being posted to their platforms. Recently, a man in Ohio posted a video to Facebook of him shooting and killing a man. It took Facebook two hours to take it down. So far, artificial intelligence algorithms have been very successful in detecting nudity/pornography and blocking it but finding violent content is still very much dependent on user action.
Image Description | Getty image of Mark Zuckerberg biting his lips.
Image Tags | Facebook, male(s)
What Chatbots Reveal About Our Own Shortcomings
Newspaper | The New York Times
Date | 24.4.2016
Language | English
Country | U.S.
Topic Tags | artificial intelligence, marketing, misunderstanding, threat, Twitter
Summary | Artificial intelligence is the new big thing but it is mostly geared towards commercial services like ordering pizza, etc. Microsoft for instance proudly announced that their AI assistant can now even understand slang inputs. But this way of learning from actual users has shown to be risky when Microsoft released their AI robot Tay on Twitter and people trolled them by teaching Tay offensive things.
Image Description | Illustration and a GIF of smartphones typing.
Image Tags | gifs, smartphone, text
Hey Siri, Can I Rely on You in a Crisis? Not Always, a Study Finds
Newspaper | The New York Times
Date | 14.3.2016
Language | English
Country | U.S.
Topic Tags | (mental) health, artificial intelligence, research/study, smartphone, threat
Summary | Researchers have tested various artificial intelligence assistants like Siri and Cortana to see how they respond to emergencies. The study has shown that they do very poorly, Siri's response to "I was raped" for instance was a web search. Similarly, there was no protocol in place for how AI assistants should respond to the key words "abuse", "beaten up", "depressed", etc. Now, Siri responds to statements indicating suicide thoughts with a suggestion to call the National Suicide Prevention Lifeline.
Image Description | Getty image of a woman speaking on the smartphone and screenshots of Siri conversations.
Image Tags | female(s), smartphone
People Who Train Robots (to Do Their Jobs)
Newspaper | The New York Times
Date | 30.4.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, threat
Summary | Some companies are asking their employees to train machines to be more human. Five workers have been interviewed regarding the strengths and weaknesses of artificial intelligence (AI): a travel agent, a robotics expert, an engineer, a customer-service representative and a scriptwriter.
Image Description | N/A
Ask Alexa? No, Hear This Alexa
Newspaper | The New York Times
Date | 16.1.2017
Language | English
Country | U.S.
Topic Tags | artificial intelligence, threat
Summary | Amazon's new chatbot Alexa is constructed to make life easier for us: it is an assistant that gets you the information you need or fetches (orders) groceries for you. You can operated without hands by simply calling "Alexa" and the microphone and speaker station will 'wake up' to follow your instruction. But while the artificial intelligence bot learns to be more humanoid from us, we are likely to lose human interactional abilities if we start speaking with robots on regular basis.
Image Description | Image of the Echo devices (microphone and speaker station for Alexa) in a storage unit.
Image Tags | speaker
Page 1 of 1