Group items tagged

"When Conor Sprouls, a customer service representative in the call center of the insurance giant MetLife talks to a customer over the phone, he keeps one eye on the bottom-right corner of his screen. There, in a little blue box, A.I. tells him how he's doing. Talking too fast? The program flashes an icon of a speedometer, indicating that he should slow down. Sound sleepy? The software displays an "energy cue," with a picture of a coffee cup. Not empathetic enough? A heart icon pops up."

La graal de la voix selon Amazon : avoir une conversation sans y avoir pensé en terme de programmation de bout en bout. Un film- the night out scenario présenté lors de cette rencontre montre une conversation fluide avec alexa pour commander et réserver plusieurs choses : une place de cinéma, un diner et un taxi. L'objectif est de faire accomplir par l'assistant vocal des taches complexes incorporant plusieurs "skills" et permettant de réduire le nombre d'itérations pour faire des tâches comme réserver une place de cinéma ou commander à manger. La voix définira les contours de la technologie des 10 prochaines années. Je veux utiliser le même assistant sur tous les supports (devices) et non par support : l'assistant de la voiture, l'assistant de la télé,...

"With "Project Alias," Bjørn Karmann and Tore Knudsen of Denmark demonstrated a simple yet effective way to take back control over our own private sphere, which earned them the STARTS Prize of the European Commission." "None of us had any smart home devices for a long time, and we were certainly not planning to buy one either. But after Bjørn won Google's AI Experiment Challenge back in 2017 he got a Google Home smart device as a gift from the I/O event. Despite the lack of excitement, Bjørn brought it home for a trial period. It did not take long before frustrations started to build up, and empathy towards the uncanny voice disappeared. Every interaction and behavioral pattern felt predetermined. He felt like a passive consumer who just wanted to give the poor assistant a name that was not the company's brand. And then there was the microphone. A direct link to the servers at Google that always was on and ready to be triggered by your command. So the idea of hacking it started to arise. But as makers, we had to face the sad truth that it was yet another device with a completely closed system. To hack it, we would have to get creative. This is when the idea of a "man-in-the-middle-attack" started. A device that we would trust and which has no connection to the internet. Whose job is to take control of the other assistant and whose name it would deserve."

"Pour ne plus dépendre des solutions fournies par les Gafam et les Batx, les acteurs français du commerce vocal s'associent au sein du Voice lab pour créer notamment un tronc commun de 100 000 heures de données vocales en français sur lesquelles entraîner leurs technologies vocales. Des start-up (Voxist, Snips, Smartly.AI, batvoice...), des éditeurs de logiciels (Linagora…) et des institutionnels (LIMSI, Université Paris Dauphine….) ont déjà rejoint l'initiative."

"This study investigated the speech recognition abilities of popular voice assistants when being verbally asked about commonly dispensed medications by a variety of participants. " "Overall, these findings demonstrated that Google Assistant possesses a much more advanced AI system than its voice assistant competitors when comprehending medication names, but there is still room for improvement. Voice assistants and proper speech recognition in healthcare has the potential to deliver efficient and important health information to patients, especially to those with reduced ability to read small font of medication labels or type on a mobile device.13 However, the presence of a human transcriber for quality assurance is still vital when assessing something as important as the information related to complex medications.14 Future advancements of this technology are critical if it should be used to achieve the comprehension levels of a human healthcare practitioner."

"- 34% of U.S. 18-29-year-olds owned a smart speaker at the beginning of 2019 while the figure for the over 60 population was just 20%. - The use of voice assistants on smartphones follows a similar pattern but is less extreme with over 80% of under 30 consumers saying they've used a smartphone-based voice assistant compared to about 60% for the oldest age group. - However, the frequency of smart speaker use is less age-dependent as 60 years and older device owners actually exceed that of the 18-29 group with 46.6% as daily users compared to 43.1% respectively. The 30-44 age group was most active with 51.3% daily users. - Consumers over 60 skew significantly toward Amazon Alexa use for smart speakers and Apple Siri on iPhones. Google Assistant and its primary distribution of Google Home smart speakers and Android smartphones are not very popular among older Americans."

"Most often, they ask them to play their favorite songs, tell jokes, and sometimes, to do their homework. Of course, chatting with a bot comes with privacy concerns." "Those concerns have created a market for voice assistants that are more kid-friendly and less nosy, such as Chatterbox, a DIY AI for kids, and Toniebox, an entertainment-focused smart speaker that sells in privacy-friendly Europe."

"The AI sausage that voice technology relies on gets made in a feedback loop: The products perform well enough, voice data from customers are collected and used to improve the service, more people buy in to the product as it improves, and then more data are collected, improving it further. This loop requires a large customer base to sustain itself, which raises the question: Would as many people have bought these products if they knew that Romanian contract workers would listen to them, even if they didn't deliberately trigger their devices? " "In a recent article for The New Yorker on the risks of automation, the Harvard professor Jonathan Zittrain coined the phrase intellectual debt: the phenomenon by which we readily accept new technology into our lives, only bothering to learn how it works after the fact. Essentially, buy first, ask questions later." "But privacy alone won't solve this. It's just as important for consumers to know how our devices work to begin with."

"The ad is called Morning Ritual which has both a 30 second and minute-long version. Created by ad agency Joint London, the ad walks through a woman's morning routine which includes frequent interactions with the Alexa voice assistant. The viewer only eventually notices the fact that the woman is unable to see, as she faces a rain-soaked window and asks Alexa what the weather is like. The whole visual impairment experience is enhanced by the low-lighting throughout the video."

"Improvements in software technology have made automatic speech transcription possible. By capturing vast quantities of human speech, neural network programs can be trained to recognize spoken language with accuracy rates that in the best circumstances approach 95 percent. Coupled with the plunging cost of storing data, it is now possible to use human language in ways that were unthinkable just a few years ago." "Mr. Liang, a Stanford-educated electrical engineer who was a member of the original team that designed Google Maps, said that data compression had made it possible to capture the speech conversation of a person's entire life in just two terabytes of information - compact enough to fit on storage devices that cost less than $50."

"Parental supervision via voice assistant won't be limited to what happens at home. Amazon announced that it will roll out new Alexa education skills in partnership with education tech companies like Blackboard, Canvas, and Coursera. Students, and their parents, will shortly be able to ask Alexa about homework or for updates posted by teachers on those platforms to stay up to date on classroom activities." "Amazon wants to increase its market share among children, and make them loyal to the voice assistant at an early age. But with children, parents are the gatekeepers. It remains to be seen if the family-friendly features and increased transparency are enough for parents to trust Alexa interacting with their children after the many privacy breaches. If successful, it could set the template for how voice assistant developers approach the market for the younger audience. Still, there's no guarantee any of it will be enough to appease worried parents or head off the kind of regulation Amazon and other voice assistant makers want to limit."

"In a recent paper that made waves in the ML community, a team at the University of Massachusetts, Amherst, found that training a model for natural-language processing - the field that helps "virtual assistants" like Alexa understand what you're saying - can emit as much as 626,155lb of carbon dioxide. That's about the same amount produced by flying roundtrip between New York and Beijing 125 times." "When it comes to ML, a group of researchers are calling for a more energy-conscious approach, which they call "Green AI". "

"Patients can tell Alexa to ask for specific help, such as pain medication or assistance using the bathroom. The skill will then send their request to their nurse via a mobile device and a tablet at the nurse's station. For patients unable to find or reach the standard call button, a voice-activated system can be a real boon. And, because the specifics of the request are shared, the skill is more efficient than a call button. Nurses will know what the patient needs before they arrive and its severity - if it's an emergency or a less life-threatening need. "

"[...]Alexa Conversations, a new deep learning-based approach that developers can use for creating natural voice experiences on Alexa with less effort, fewer lines of code, and less training data than before. This new model helps developers create natural, flexible dialogs within a single skill and in the upcoming releases brings multiple skills into a single conversation. Alexa Conversations is now available in developer preview in the US."

"[...] fears of a dystopian AI future are also swirling around Amazon's facial recognition software system Rekognition, which the company reportedly attempted to sell to U.S Immigration and Customs Enforcement (ICE) and which has been tested by law enforcement in Washington and Oregon. While the public is currently focused on the use of Rekognition by law enforcement and government agencies, there is another issue to consider. If the technology becomes a standard part of the smart display experience, people may balk at the idea of installing Amazon's technology in their homes."

"In a free service, bots call restaurants and make reservations. The technology is impressive, except for when the caller is actually a person." "In other words, Duplex, which Google first showed off last year as a technological marvel using A.I., is still largely operated by humans. While A.I. services like Google's are meant to help us, their part-machine, part-human approach could contribute to a mounting problem: the struggle to decipher the real from the fake, from bogus reviews and online disinformation to bots posing as people." "Google's A.I. is eerily human, when it works" "Duplex needs lots of data to improve" "Duplex is proficient at making a restaurant reservation over the phone, but much like Facebook, Google still leans on human intelligence. At any given moment, it is lifelike. But it struggles to deal with the unexpected. "There are three things that are important when it comes to A.I.'s interactions with humans: context, context and context," said Jerry Kaplan, author of "Humans Need Not Apply: A Guide to Wealth and Work in the Age of Artificial Intelligence" and a Stanford University lecturer on artificial intelligence. "Machines are very good with detail but terrible at context," he said."