Number of Posts: 9
Posts 1 - 9
SmileTracker, l'application qui épie vos émotions
(SmileTracker, the app that is spying on your emotions)
Newspaper | Le Monde
Date | 27.8.2017
Language | French
Country | France
Topic Tags | artificial intelligence, smartphone
Summary | SmileTracker is an app that can detect smiles thanks to a face recognition system. Users can gather all their pictures in one "smile gallery" and then share them on the SmileTracker website or Facebook. If people see those pictures, it can trigger positive emotions and a general well-being. This SmileTracker app is one of MIT's experiments. Researchers at MIT want to create machines capable of recognizing and measuring human emotions.
Image Description | N/A
Au-delà des fantasmes, quels sont les problèmes concrets que pose l'intelligence artificielle?
(Beyond fantasy, what are the real problems that poses artificial intelligence?)
Newspaper | Le Monde
Date | 4.8.2017
Language | French
Country | France
Topic Tags | artificial intelligence, threat
Summary | When Elon Musk said that robots were killing people and that we should be worried about that, Mark Zuckerberg replied that Musk's ideas were irresponsible and also praised the advances of artificial intelligence. Even if Musk's pessimism seems exaggerated, artificial intelligence raises some concerns and problems (e.g. impact on jobs, racist and sexist softwares, creation of autonomous weapons, surveillance problems, etc.)
Image Description | N/A
Non, Facebook n'a pas «paniqué» à cause d'un programme d'IA capable d'inventer un langage
(No, Facebook didn't "panick" because of an AI program capable of inventing a language)
Newspaper | Le Monde
Date | 1.8.2017
Language | French
Country | France
Topic Tags | artificial intelligence, Facebook, threat
Summary | Dhruv Batra, a researcher in artificial intelligence, read many Facebook posts and tweets describing apocalyptic scenes about the fact that artificial intelligence had invented a new language. Batra wrote an article denouncing how the media portrayed the event. When Facebook programmed two bots to teach them how to negotiate, employees noticed that the bots had invented their own language. The new language was based on English vocabulary; therefore, it was more or less comprehensible. But since the program was not working the way it was supposed to work, workers modified it. They didn't "kill" it, so there is no reason to panick.
Image Description | N/A
Facebook: «Nous voulons faire d'Internet une “no-go zone" pour les terroristes»
(Facebook: "We want to make the Internet a "no-go zone" for terrorists)
Newspaper | Le Monde
Date | 16.6.2017
Language | French
Country | France
Topic Tags | artificial intelligence, Facebook, threat
Summary | Facebook just issued a statement where they explain how they want to fight propaganda and terrorism. Monika Bickert and Brian Fishman talk about why they released the statement now, the technology used to fight terrorism, artificial intelligence, privacy, and threatening content.
Image Description | N/A
L'algorithme qui comprend les contes pour enfants
(The algorithm that understands children's tales)
Newspaper | Le Monde
Date | 28.2.2017
Language | French
Country | France
Topic Tags | artificial intelligence, Facebook, research/study
Summary | One of Facebook's research labs on artificial intelligence created an algorithm capable of understanding and remembering texts. The algorithm is even better than the "tale's text". The essence of intelligence is the capacity to predict, which is also one of the goals of artificial intelligence. Chatbots don't understand the questions that people ask; they actually react to keywords.
Image Description | N/A
A la rencontre de Watson, l'intelligence artificielle star d'IBM
(Meet Watson, IBM's artificial intelligence star)
Newspaper | Le Monde
Date | 18.3.2017
Language | French
Country | France
Topic Tags | artificial intelligence
Summary | Watson is an artificial intelligence program that can analyze your psychological profile while chatting with you. Watson can for example suggest a beer that would be adapted to your taste based on a series of questions. The program does not only recommend beers, it also gets involved in hospitals and banks. Watson can recognize words, images, and can analyze people's tone and emotions. The goal is not to create people's replicas, but to find out what people's limitations are.
Image Description | N/A
L'intelligence artificielle reproduit aussi le sexisme et le racisme des humains
(Artificial intelligence also reproduces human beings' sexism and racism)
Newspaper | Le Monde
Date | 15.4.2017
Language | French
Country | France
Topic Tags | artificial intelligence, gender, research/study, threat
Summary | Gender stereotypes are reproduced in some artificial intelligence programs. Researchers at the University of Stanford show how machine learning can replicate people's biases. They based their research on a technology called GloVe, which is trained to look for common associations. The technology points to some problematic associations that illustrate sexism and racism. The fact that AI follows people's prejudices can have some serious consequences, so people are trying to find solutions against AI's biases.
Image Description | N/A
La «dictée par la pensée» sur laquelle travaille Facebook n'est pas encore pour demain
(The "mind dictation", on which Facebook is working, is not for tomorrow)
Newspaper | Le Monde
Date | 17.4.2017
Language | French
Country | France
Topic Tags | artificial intelligence, brain, Facebook
Summary | At a Facebook conference on April 19th, it was announced that Facebook is working on a new project that should allow people to type directly from their mind/thoughts. Researchers want to decode language in order to allow people to type their thoughts. The goal would be to type 100 words per minute based on a person's thoughts. Decoding language directly from the brain is difficult. Moreover, Facebook is facing a major difficulty with the sensor technology they want to develop. Finally, besides technical difficulties, there are also ethical ones.
Image Description | N/A
Malgré d'impressionnants progrès, la traduction automatique a encore du chemin à parcourir
(Despite tremendous progress, machine translation still has a long way to go)
Newspaper | Le Monde
Date | 19.5.2017
Language | French
Country | France
Topic Tags | artificial intelligence, translation
Summary | Facebook, TripAdvisor, and Google Chrome often ask users if they want to translate their content into French. Translations have improved, but translated texts can still be imperfect and odd. Still, translated texts are usually good enough to allow people to understand original texts. Facebook, Google, and Microsoft are still trying to improve their translation tools to make them faster and more accurate.
Image Description | N/A
Page 1 of 1