“Gemini is built from the ground up for multimodality — reasoning seamlessly across text, images, video, audio, and code.”
Source : Gemini – Google DeepMind
Étiquette : generative ai (Page 1 of 2)
“In this short video, hear from Google leaders and AI experts as they introduce you to Gemini — Google’s largest and most capable AI model. It’s built from the ground up to be multimodal — meaning that it’s trained to recognize, understand and combine different types of information, including text, images, audio, video and code. And it’s optimized in three different sizes: Ultra, Pro and Nano. Welcome to the Gemini era.”
via Google
“After Google quietly scrapped a set of in-person events to launch Gemini, its biggest artificial intelligence initiative in a decade, the company has planned a virtual preview of the new AI as soon as this week, said a person with knowledge of the situation. By giving journalists and software developers a first look at some of the technology’s capabilities, Google could relieve some pressure from investors to prove it can catch up to ChatGPT creator OpenAI. Google representatives for weeks have been giving private demonstrations of the technology to business partners but they have said that cloud customers wouldn’t get access to the primary version of Gemini until next year.”
Source : Google Preps Public Preview of Gemini AI After Postponing In-Person Launch Events — The Information
We ran a series of experiments to measure the energy efficiency and carbon emissions of different models from the HuggingFace Hub, and to see how different tasks and models compare. We found that multi-purpose, generative models are orders of magnitude more energy-intensive than task-specific systems for a variety of tasks, even for models with a similar number of parameters
“Beyond detailing the threat posed by generative AI tools that can tweak images, Tech Against Terrorism has published a new report citing other ways in which gen AI tools can be used to help extremist groups. These include the use of autotranslation tools that can quickly and easily convert propaganda into multiple languages, or the ability to create personalized messages at scale to facilitate recruitment efforts online. But Hadley believes that AI also provides an opportunity to get ahead of extremist groups and use the technology to preempt what they will use it for.
»We’re going to partner with Microsoft to figure out if there are ways using our archive of material to create a sort of gen AI detection system in order to counter the emerging threat that gen AI will be used for terrorist content at scale,” Hadley says. “We’re confident that gen AI can be used to defend against hostile uses of gen AI. »”
Source : Here’s How Violent Extremists Are Exploiting Generative AI Tools | WIRED
“Artificial intelligence image tools have a tendency to spin up disturbing clichés: Asian women are hypersexual. Africans are primitive. Europeans are worldly. Leaders are men. Prisoners are Black.
These stereotypes don’t reflect the real world; they stem from the data that trains the technology. Grabbed from the internet, these troves can be toxic — rife with pornography, misogyny, violence and bigotry.”
Source : AI generated images are biased, showing the world through stereotypes – Washington Post
“To the human eye, the Glazed image still looks like her work, but the computer-learning model would pick up on something very different. It’s similar to a tool the University of Chicago team previously created to protect photos from facial recognition systems.
When Ms. Ortiz posted her Glazed work online, an image generator trained on those images wouldn’t be able to mimic her work. A prompt with her name would instead lead to images in some hybridized style of her works and Pollock’s.
“We’re taking our consent back,” Ms. Ortiz said. A.I.-generating tools, many of which charge users a fee to generate images, “have data that doesn’t belong to them,” she said. “That data is my artwork, that’s my life. It feels like my identity.”
The team at the University of Chicago admitted that their tool does not guarantee protection and could lead to countermeasures by anyone committed to emulating a particular artist. “We’re pragmatists,” Professor Zhao said. “We recognize the likely long delay before law and regulations and policies catch up. This is to fill that void.””
Source : This Tool Could Protect Artists From A.I. Image Generators – The New York Times
“This Spotify-developed tool leverages the latest innovations—one of which is OpenAI’s newly released voice generation technology—to match the original speaker’s style, making for a more authentic listening experience that sounds more personal and natural than traditional dubbing. A podcast episode originally recorded in English can now be available in other languages while keeping the speaker’s distinctive speech characteristics. ”
“We are beginning to roll out new voice and image capabilities in ChatGPT. They offer a new, more intuitive type of interface by allowing you to have a voice conversation or show ChatGPT what you’re talking about.”
Source : ChatGPT can now see, hear, and speak
“Les enseignant·es sont souvent sous l’eau et n’ont pas le temps de se former tout le temps. Nous devons évidemment leur fournir des outils et proposer des formations. Mais à la vitesse à laquelle évoluent les outils tels que ChatGPT et le temps que prennent parfois les formations, force est de constater qu’il y a un problème. Nous devons trouver des solutions pour ne pas cesser de former les enseignant·es, en continu et de la façon la plus efficace possible sur les sujets qui touchent de façon générale au numérique. Ce qui pose des questions nombreuses et complexes, en termes de formats, de quantité de matière et de pertinence selon les disciplines. Un enjeu plus facile à relever, sur lequel nous travaillons déjà, est la formation sur les connaissances de base en sciences informatiques.”