Category: Artificial intelligence

  • Arc to Dia: New AI Web Browser Plans from The Browser Company

    Arc to Dia: New AI Web Browser Plans from The Browser Company

    Arc, the beloved Chromium browser known for its innovative handling of tabs and organization, is no longer the center of excitement for Josh Miller, the CEO of The Browser Company. Instead, the startup has shifted its focus to Dia, a more mainstream product that leverages AI to simplify web browsing. Set to debut in early 2025, Dia is not just an upgrade of Arc; it’s meant to be a complete replacement.

    Features Unveiled

    In a recent presentation cleverly disguised as a recruiting video, Dia showcases three prototype demonstrations that hint at potential features of its new ‘computing environment’. One standout feature transforms the traditional insertion cursor into an AI-driven tool that can suggest text to help you “write the next line” and more. This personalized cursor can analyze the whole browser window rather than just a single text box. For instance, it can take prompts to copy Amazon links from open tabs and seamlessly add them into an email draft.

    Smart Searching

    In another demonstration by Josh Miller, he uses the address bar to locate a document by inputting only the sender and the theme. Dia efficiently locates the exact Notion document and, upon request, sends it via the chosen email client. These features utilize natural-language processing capabilities from LLMs, along with memory and autonomous actions that should be inherent to the browser.

    Automating Tasks

    The third prototype reveals Dia’s ability to automatically add items from an email to an Amazon shopping cart – a rather complicated AI challenge. In yet another demo, the browser uses a template to send personalized emails to each team member based on a list with their specific call times. In both instances, Miller suggests that the browser can be trained to handle such repetitive tasks with ease.

    Miller appears to believe that AI functionalities like these will turn the simple browser into a robust operating system through this new computing environment. Understandably, users who have been loyal to Arc are not taking this news lightly. However, if the recent updates and bug fixes are any indication, Arc might still receive some support, though new features are unlikely. With Dia on the way, The Browser Company has fully embraced AI advancements, and there’s no turning back now.

    Source: Link

    Tags: , ,
  • Canadian News Outlets Sue OpenAI for Copyright Violations

    Canadian News Outlets Sue OpenAI for Copyright Violations

    A group of Canadian news organizations has filed a joint lawsuit against OpenAI, claiming that the company has unlawfully utilized their articles to develop its AI model.

    Details of the Lawsuit

    According to CBC, the lawsuit includes prominent news outlets like CBC/Radio-Canada, Postmedia, Metroland, the Toronto Star, the Globe and Mail, and the Canadian Press. They have submitted their case to the Ontario Superior Court of Justice, seeking “punitive damages” along with “payment of any profits that the company made from using news articles from the organizations.”

    Ongoing Legal Battles

    This lawsuit is just one of many that news outlets have brought against OpenAI, accusing the company of scraping their data to enhance its chatbots. OpenAI is currently facing lawsuits from The New York Times, Center for Investigative Journalism, Chicago Tribune, ANI, and a host of other organizations.

    OpenAI’s Response

    In a conversation with The Verge, Jason Deutrom, a spokesperson for OpenAI, mentioned that the company works “closely with news publishers,” ensuring proper display, attribution, and linking to their content in ChatGPT search. He also stated that they provide simple methods for publishers to opt out if they wish.

    In a collective statement, the Canadian news organizations asserted that OpenAI is “capitalizing and profiting” from their content “without obtaining permission or compensating content owners.”

    In an email to CBC, Deutrom explained that OpenAI’s models were trained on data that is publicly accessible, affirming that the company follows “international copyright principles.”

    Tags: , ,
  • OpenAI’s Sora Video Generator Leaked on Hugging Face

    OpenAI’s Sora Video Generator Leaked on Hugging Face

    OpenAI revealed its new text-to-video generation model named Sora back in February. The company stated that Sora could create videos that last up to a minute and manage complex scenes involving multiple characters and movements.

    Anonymous Sharing of Sora

    Even though OpenAI hasn’t officially confirmed when it will be available to the public, a group of anonymous artists has allegedly shared the model on the Hugging Face machine learning platform. According to Techcrunch, the model was accessible for about three hours until OpenAI took it down.

    Limited Access for Users

    Not everyone was able to beat the long lines to create videos, but some users did manage to upload examples that were similar to what OpenAI had showcased. The group that leaked the model claimed they felt pressured to "inform the world that Sora is an effective tool for artists."

    Artists Voice Their Concerns

    After the leak, the group alleged that OpenAI quickly restricted access to all artists. In an open letter, they stated, "While hundreds are contributing for free, only a select few will be picked through a competition to have their films created with Sora shown — providing very little compensation that is insignificant compared to the large PR and marketing benefits OpenAI gains."

    The letter expressed the group’s desire to "denormalize" the practice of using artists as unpaid PR and R&D for "billion-dollar companies." They also urged individuals to sign a petition requesting that major brands "create a genuine path for artist expression, offering fair compensation to the creators."

    Source: Link,Link,Link,Link,

    Tags: , ,
  • Runway AI Launches New Image Generation Model: Frames

    Runway AI Launches New Image Generation Model: Frames

    Runway AI, an AI research and tech company based in New York, has launched a new image creation model called Frames, which signifies “a significant advancement in stylistic precision and visual quality.”

    Focus on Stylistic Consistency

    This model seems to focus on keeping a consistent style, meaning it can create similar images while adding or removing certain details. This is an area where many other image generation models have difficulty.

    Runway claims that with Frames, users can select a style for their project “and consistently produce variations that align with your aesthetic.”

    Showcasing Stylistic Choices

    Runway presented some examples showcasing stylistic consistency with Frames on an X thread. The styles, referred to as “Worlds,” include a range from 1980s special effects, 1970s album cover art, Japanese zines, realistic digital images, landscapes, and much more.

    The generated images appear striking at first sight and keep a uniform theme and vision. Runway mentioned that it will slowly provide access to Frames within the Gen-3 Alpha of its video creation suite and Runway API.

    Applications in Media

    Runway’s video generation tools have been used in movies like Everything Everywhere All At Once, and in music videos for artists such as Kanye West. They’ve also found their way into popular TV shows like Top Gear.

    Source: Link,Link

    Tags: , ,
  • Nvidia Hopper and Blackwell Drive Strong Q3 Revenue Growth

    Nvidia Hopper and Blackwell Drive Strong Q3 Revenue Growth

    Nvidia is setting new financial records, as shown in its latest quarterly report, nearly doubling its revenue from the third quarter of last year (2024). The tech giant announced a remarkable performance for the three months that ended in October 2024, bringing in revenue of $35.1 billion—a 17% increase from the last quarter and an incredible 94% rise compared to the same time last year.

    Data Center Growth

    The report indicates that a large part of this growth came from Nvidia’s data center sector, which achieved record revenue of $30.8 billion. This marks a 17% increase from the previous quarter and a stunning 112% rise compared to the same quarter last year. A key factor in this success was the high demand for specific chips like the Hopper and Blackwell platforms, as customers are increasingly using Nvidia-powered systems for a variety of applications, from training large language models to supporting AI-driven cloud services.

    Cloud Partnerships

    Nvidia also pointed out the strengthening of its global cloud partnerships, with major players like Amazon Web Services (AWS), Microsoft Azure, and CoreWeave providing instances powered by the Hopper H200. The company feels that this collaboration has played a crucial role in maintaining its lead in the AI infrastructure market.

    Gaming and Professional Visualization

    Moreover, the report highlighted growth in Nvidia’s gaming and professional visualization sectors, although the increases were less significant. For Q3, the gaming revenue reached $3.3 billion, which is a 14% rise from the previous quarter. This growth is linked to the ongoing transition in gaming towards AI-enhanced graphics and immersive experiences. New AI-powered PCs and games, like Indiana Jones and the Great Circle, are pushing the limits of interactive entertainment.

    The revenue from the professional visualization segment saw a 17% increase year-over-year, as various industries are adopting Nvidias’s Omniverse platform to develop digital twins and optimize workflows in manufacturing, media, and design.

    Future Projections

    Looking forward, the company anticipates that the momentum will carry on into the fourth quarter, projecting a revenue of $37.5 billion, which reflects the continuing global demand for AI infrastructure and solutions.

    Source: Link

    Tags: , ,
  • OpenAI Accused of Deleting Evidence in Copyright Lawsuit

    OpenAI Accused of Deleting Evidence in Copyright Lawsuit

    In recent legal documents submitted by The New York Times regarding its lawsuit against OpenAI, it has been claimed that the AI company unintentionally erased crucial evidence. This evidence was gathered by the newspaper’s legal team over a span of more than 150 hours.

    Data Recovery Issues

    The documents indicate that OpenAI was able to retrieve most of the data, however, the folder organization and file names were irretrievably lost. Consequently, the data that was recovered does not allow for an understanding of how the news articles from the plaintiffs were incorporated into the training sets for the AI model.

    OpenAI’s Response

    Jason Deutrom, a representative from OpenAI, communicated to Wired that the company does not agree with the claims made in the lawsuit. He mentioned that they will be providing a formal response shortly. The New York Times initiated the lawsuit against OpenAI and Microsoft in December of the previous year, claiming that the AI firm utilized its articles to enhance its AI models.

    Accusations of Copyright Infringement

    The lawsuit asserts that millions of articles from The New York Times were used to train automated chatbots, which now rival the news organization as a trusted source of information. It argues that both companies should be held liable for "billions of dollars" due to the "illegal copying and use of The Times’ uniquely valuable works."

    Source: Link,Link,Link,Link

    Tags: , ,
  • Adobe SlimLM Offers Cloud-Free Mobile AI Solutions

    Adobe SlimLM Offers Cloud-Free Mobile AI Solutions

    Thanks to a groundbreaking small language model (SLM) created by a group of AI researchers at Adobe Inc. working with teams from Auburn University and Georgia Tech, new smartphone AI applications will now be able to handle documents directly on the device without needing to connect to the cloud. This model is named SlimLM, and all relevant information about it has been shared on arXiv.

    Efficient On-Device Processing

    “Through thorough experiments on a Samsung Galaxy S24, we pinpoint the best trade-offs between the model’s size (which varies from 125M to 7B parameters), context length, and inference time for effective processing on the device,” shared the research group. “SlimLM is pre-trained on SlimPajama-627B and fine-tuned on DocAssist, our specially created dataset for tasks like summarization, question answering, and suggestions. Our smallest model shows great performance on the S24, while the bigger versions provide enhanced features within the limits of mobile devices.”

    Future Availability

    SlimLM isn’t available to the public just yet, but that could change soon. With this innovative code, smartphones can harness AI’s power locally to process documents without needing internet access. While cloud-based AI solutions typically rely on less local computing power compared to apps that run directly on devices, the key benefit they provide is privacy. Major players in this space, like Google, Apple, and Meta, have already launched similar applications, but their status is largely experimental, and they haven’t reached widespread use. If the goals of these researchers are successful, SlimLM could potentially be the first widely adopted solution of its kind.

    Source: Link

    Tags: , ,
  • Samsung Unveils Second-Generation AI Model Gauss2

    Samsung Unveils Second-Generation AI Model Gauss2

    Samsung presented the next version of its multimodal generative AI model, Gauss2, during the Samsung Developer Conference 2024. The model will come in three different sizes: Compact, Balanced, and Supreme.

    Compact Model Overview

    The Compact model is designed to be small, focusing on speed and efficiency, making it perfect for usage directly on devices by "maximizing the utilization of the device’s computing resources."

    Balanced and Supreme Models

    The Balanced model strikes a blend between performance and efficiency, tailored for various tasks that need consistency. On the other hand, the Supreme model is optimized for high performance, and Samsung claims it lowers "computational costs during training and inference processes while keeping both performance and efficiency at high levels."

    This model supports up to 14 languages, including a variety of programming languages. To enhance efficiency and performance, Samsung employs a "custom tokenizer" for the languages it supports. The company notes that the model’s processing speed "per hour is 1.5 to 3 times quicker," when compared to open-source generative AI models.

    In-House Coding Assistant

    According to Samsung, its internal coding assistant, ‘code.i,’ is currently being used by 60% of all software developers at the company.

    Samsung aims to leverage the Gauss2 model to boost productivity internally. The company "will keep extending the reach of its AI-based services throughout all product lines so users can enjoy a more convenient and pleasant daily life."

    Source: Link

    Tags: , ,
  • Free AI Training Course for Teachers by OpenAI

    Free AI Training Course for Teachers by OpenAI

    OpenAI has joined forces with Common Sense Media to provide a complimentary course for educators, aiming to enhance their understanding of generative AI. Common Sense Media is a nonprofit that evaluates technology suitable for kids and families.

    Course Details

    According to a report by Reuters, the course named "ChatGPT Foundations for K-12 Educators" will explore effective practices for using AI and prompt engineering. This course is primarily focused on ChatGPT and will assist educators in learning the fundamentals of the chatbot, alongside showcasing methods for incorporating it into their teaching techniques within the classroom.

    Target Audience

    The course is designed for teachers who work with students from Kindergarten to 12th Grade and is expected to be completed in under an hour. It includes key topics such as the essentials of AI, responsible usage, data privacy, and practical suggestions for integrating AI into educational environments. OpenAI mentioned that the course was tested in around twelve school districts, where 98% of the educators reported gaining new insights for their teaching.

    Expert Insight

    Leah Belsky, who serves as the VP and General Manager of Education at OpenAI, shared with Reuters that the course aims to "put AI into the hands of every student and every teacher… and also equip them with the skills to learn how to do it responsibly and effectively." This initiative represents a significant step toward making AI tools more accessible for educational purposes.

    Source: Link,Link

    Tags: , ,
  • Amazon Delays Alexa’s AI Update Due to Latency Issues

    Amazon Delays Alexa’s AI Update Due to Latency Issues

    Leaked internal documents obtained by Fortune reveal various technical issues related to the upcoming AI revamp for Alexa, Amazon’s voice assistant. A major issue is latency, which refers to the delay in how long it takes for the AI to answer a question.

    Testing Feedback and Concerns

    The latency concern is serious enough that several internal testers have rated the new assistant poorly. The average score given by testers was 4.5 out of 7, falling short of the target score of 5.5 out of 7. The team is worried that if this product is launched in its current condition, it could lead to user frustration.

    Key Issues Highlighted

    The memo states that “latency continues to be a major issue” that will need “notable improvements.” Another significant challenge is the possibility that some older Echo devices might not work with the new assistant.

    Amazon spokesperson Lauren Raemhild mentioned to Fortune, “Generative AI presents a significant opportunity to enhance Alexa for our customers.” She added, “We are diligently working to provide even more proactive and capable assistance on the more than half a billion Alexa-enabled devices already present in homes globally.”

    Challenges Faced by the Team

    The team is also encountering numerous technological and administrative challenges, such as insufficient data for training its AI model or resources being redirected to focus on Generative AI for Amazon’s cloud services. The memos indicate that the relaunch of Alexa has been postponed at least three times, with the latest date set for November 14, 2024.

    Source: Link

    Tags: , ,