Tag • #microsoft

chevron_right

Mikerosoft – Oups la boulette chez Microsoft

news.movim.eu / Korben · Yesterday - 16:46 · 2 minutes

Imaginez le bazar si quelqu’un ajoutait par erreur tous les Jean-Michel de votre entreprise dans une conversation de groupe ! C’est un peu ce qui est arrivé chez Microsoft cette semaine. De nombreux employés prénommés Mike ou Michael ont eu la surprise de se retrouver sans le vouloir dans une boucle d’échanges par mail.

Michael Schechter , le VP de Bing , raconte s’être réveillé avec une quantité inhabituelle d’emails non lus. Sur le coup, il a cru à un gros plantage en prod pendant la nuit, mais non, en fait c’est juste une personne qui s’est amusée à créer un groupe avec beaucoup de gens qui s’appellent Mike ou Michael chez Microsoft. Et pas de bol pour eux, ils sont nombreux !

Face à cette situation ubuesque, Michael a eu le réflexe de demander à Copilot (l’assistant d’IA de krosoft) de résumer le fil de discussion et croyez-moi, ça vaut son pesant de cacahuètes ! D’après Copilot, les participants ont commencé par demander des explications sur le but de ce groupe, tout en notant avec amusement qu’ils avaient le même prénom. Évidemment, ça n’a pas manqué de partir en vrille et chacun y est allé de sa petite blague, de jeux de mots rigolos comme renommer le groupe en « Mikerosoft « . Certains se sont même demandé avec humour si ce n’était pas un piège pour les virer !

Le plus drôle dans l’histoire, c’est que malgré les nombreux messages, personne n’a compris qui avait créé ce groupe ni pourquoi. Un beau mystère ! En attendant, les participants en ont profité pour faire connaissance. Si ça se trouve, il y en a même qui en ont profité pour corriger des bugs… roooh.

Ça rappelle quand même qu’il faut toujours faire attention avec les emails de groupe ou les mises en copie. Une erreur est vite arrivée et on a vite fait d’envoyer des conneries et de les regretter après ! Une histoire similaire a d’ailleurs eu lieu dans les années 90 selon Eric Lippert . Un type voulait contacter « Mike de Microsoft » qu’il avait rencontré à une conf sauf qu’il a réussi à chopper les adresses des 600 Mike de la boîte. La presse avait titré à l’époque « Heureusement qu’ils ne cherchaient pas Bill » !

L’histoire ne dit pas s’il vont s’organiser un barbecue ou séminaire entre Mike, mais je troue que « Mikerosoft », ça sonne quand même mieux que Microsoft, non ?

Source

chevron_right

Microsoft’s VASA-1 can deepfake a person with one photo and one audio track

news.movim.eu / ArsTechnica · Yesterday - 13:07 · 1 minute

Enlarge / A sample image from Microsoft for "VASA-1: Lifelike Audio-Driven Talking Faces Generated in Real Time." (credit: Microsoft )

On Tuesday, Microsoft Research Asia unveiled VASA-1 , an AI model that can create a synchronized animated video of a person talking or singing from a single photo and an existing audio track. In the future, it could power virtual avatars that render locally and don't require video feeds—or allow anyone with similar tools to take a photo of a person found online and make them appear to say whatever they want.

"It paves the way for real-time engagements with lifelike avatars that emulate human conversational behaviors," reads the abstract of the accompanying research paper titled, "VASA-1: Lifelike Audio-Driven Talking Faces Generated in Real Time." It's the work of Sicheng Xu, Guojun Chen, Yu-Xiao Guo, Jiaolong Yang, Chong Li, Zhenyu Zang, Yizhong Zhang, Xin Tong, and Baining Guo.

The VASA framework (short for "Visual Affective Skills Animator") uses machine learning to analyze a static image along with a speech audio clip. It is then able to generate a realistic video with precise facial expressions, head movements, and lip-syncing to the audio. It does not clone or simulate voices (like other Microsoft research ) but relies on an existing audio input that could be specially recorded or spoken for a particular purpose.

Read 11 remaining paragraphs | Comments

chevron_right

LLMs keep leaping with Llama 3, Meta’s newest open-weights AI model

news.movim.eu / ArsTechnica · 2 days ago - 21:04 · 1 minute

A group of pink llamas on a pixelated background.

Enlarge (credit: Getty Images | Benj Edwards )

On Thursday, Meta unveiled early versions of its Llama 3 open-weights AI model that can be used to power text composition, code generation, or chatbots. It also announced that its Meta AI Assistant is now available on a website and is going to be integrated into its major social media apps, intensifying the company's efforts to position its products against other AI assistants like OpenAI's ChatGPT, Microsoft's Copilot, and Google's Gemini.

Like its predecessor, Llama 2 , Llama 3 is notable for being a freely available, open-weights large language model (LLM) provided by a major AI company. Llama 3 technically does not quality as "open source" because that term has a specific meaning in software (as we have mentioned in other coverage ), and the industry has not yet settled on terminology for AI model releases that ship either code or weights with restrictions (you can read Llama 3's license here ) or that ship without providing training data. We typically call these releases "open weights" instead.

At the moment, Llama 3 is available in two parameter sizes: 8 billion (8B) and 70 billion (70B), both of which are available as free downloads through Meta's website with a sign-up. Llama 3 comes in two versions: pre-trained (basically the raw, next-token-prediction model) and instruction-tuned (fine-tuned to follow user instructions). Each has a 8,192 token context limit.

Read 8 remaining paragraphs | Comments

chevron_right

Microsoft se souvient de l’existence des raccourcis copier-coller pour Outlook

news.movim.eu / Numerama · 3 days ago - 11:58

Microsoft va apporter une modification modeste en apparence, mais grande dans sa praticité. L'entreprise prévoit de faciliter le copier-coller des mails dans Outlook grâce aux raccourcis de copier-coller.

Sl chevron_right

Contact publication

pubsub.blastersklan.com / slashdot · 4 days ago - 19:23 edit

Microsoft's Beijing-based research group published a new open source AI model on Tuesday, only to remove it from the internet hours later after the company realized that the model hadn't gone through adequate safety testing. From a report: The team that published the model, which is comprised of China-based researchers in Microsoft Research Asia, said in a tweet on Tuesday that they "accidentally missed" the safety testing step that Microsoft requires before models can be published. Microsoft's AI policies require that before any AI models can be published, they must be approved by the company's Deployment Safety Board, which tests whether the models can carry out harmful tasks such as creating violent or disturbing content, according to an employee familiar with the process. In a now-deleted blog post, the researchers behind the model, dubbed WizardLM-2, said that it could carry out tasks like generating text, suggesting code, translating between different languages, or solving some math problems.

Read more of this story at Slashdot.

Microsoft Takes Down AI Model Published by Beijing-Based Researchers Without Adequate Safety Checks

chevron_right

Why the US government’s overreliance on Microsoft is a big problem

news.movim.eu / ArsTechnica · 4 days ago - 13:55

Enlarge (credit: Joan Cros via Getty )

When Microsoft revealed in January that foreign government hackers had once again breached its systems , the news prompted another round of recriminations about the security posture of the world’s largest tech company.

Despite the angst among policymakers, security experts, and competitors, Microsoft faced no consequences for its latest embarrassing failure. The United States government kept buying and using Microsoft products, and senior officials refused to publicly rebuke the tech giant. It was another reminder of how insulated Microsoft has become from virtually any government accountability, even as the Biden administration vows to make powerful tech firms take more responsibility for America’s cyber defense.

Read 55 remaining paragraphs | Comments

chevron_right

Windows 11 : des mises à jour de sécurité et des nouvelles fonctions à installer dès maintenant

news.movim.eu / JournalDuGeek · 6 days ago - 10:00

Windows 11

Comme chaque mois, Microsoft a procédé à la mise à jour de sécurité de ses systèmes, une pratique routinière bien connue sous le nom de Patch Tuesday. La dernière édition a toutefois attiré l'attention car elle corrige deux failles particulièrement critiques qui ont été exploitées par des acteurs malveillants (« zero day »).

chevron_right

From boom to burst, the AI bubble is only heading in one direction | John Naughton

news.movim.eu / TheGuardian · 7 days ago - 15:00 · 1 minute

No one should be surprised that artificial intelligence is following a well-worn and entirely predictable financial arc

“Are we really in an AI bubble,” asked a reader of last month’s column about the apparently unstoppable rise of Nvidia, “and how would we know?” Good question, so I asked an AI about it and was pointed to Investopedia , which is written by humans who know about this stuff. It told me that a bubble goes through five stages – rather as Elisabeth Kübler-Ross said people do with grief . For investment bubbles, the five stages are displacement, boom, euphoria, profit-taking and panic. So let’s see how this maps on to our experience so far with AI.

First, displacement. That’s easy: it was ChatGPT wot dunnit. When it appeared on 30 November 2022, the world went, well, apeshit. So, everybody realised, this was what all the muttering surrounding AI was about! And people were bewitched by the discovery that you could converse with a machine and it would talk (well, write) back to you in coherent sentences. It was like the moment in the spring of 1993 when people saw Mosaic , the first proper web browser, and suddenly the penny dropped: so this was what that “internet” thingy was for. And then Netscape had its initial public offering in August 1995, when the stock went stratospheric and the first internet bubble started to inflate .

chevron_right

Microsoft propose DALL-E au Pentagone, et les idéaux d’OpenAI s’écroulent

news.movim.eu / JournalDuGeek · Friday, 12 April - 12:59

Nadella Openai

Des documents de la Défense américaine montrent que Microsoft a suggéré au Pentagone d'utiliser le travail d'OpenAI pour améliorer ses logiciels militaires - une pente glissante qui montrent que les idéaux originaux de Sam Altman ont souffert depuis le rapprochement avec le titan du software.