Latest news with #Llama4
Economic Times
16 hours ago
- Business
- Economic Times
Mark Zuckerberg made abortive attempt to buy ex-OpenAI executive Ilya Sutskever's AI startup: Report
Meta CEO Mark Zuckerberg's $14.2 billion investment in Scale AI last week, and roping in its CEO Alexandr Wang, appears to have been part of a larger game plan to build expertise in artificial intelligence (AI). Days after Sam Altman accused Zuckerberg of trying to poach OpenAI's talent, reports have emerged that Meta has been out shopping for other startups, including Ilya Sutskever's $32 billion Safe Superintelligence, launched just a year ago. Sources told news channel CNBC that Meta tried to acquire the startup founded by Sutskever, former chief scientist at OpenAI, earlier this year. However, Sutskever rebuffed the big tech giant's Zuckerberg started eyeing the chief executive of the startup, Daniel Gross. Zuckerberg is setting up a team of experts to achieve so-called "artificial general intelligence" (AGI), or machines that can match or surpass human capabilities. The Meta chief's apparent plans to personally recruit around 50 people, including a new head of AI research for the AGI team, is driven partly by frustration over the performance and reception of Meta's latest large language model, Llama 4. Altman swiped at Zuckerberg in his brother's podcast, 'Uncapped', saying that Meta tried to poach the company's AI talent for its superintelligence team. Additionally, he said Meta is not "great at innovation". The 'special thing' about OpenAI is that the company has built a culture that is good at 'repeatable' innovation, Altman noted.
Time of India
a day ago
- Business
- Time of India
Mark Zuckerberg made abortive attempt to buy ex-OpenAI executive Ilya Sutskever's AI startup: Report
Meta CEO Mark Zuckerberg 's $14.2 billion investment in Scale AI last week , and roping in its CEO Alexandr Wang, appears to have been part of a larger game plan to build expertise in artificial intelligence (AI). Days after Sam Altman accused Zuckerberg of trying to poach OpenAI 's talent, reports have emerged that Meta has been out shopping for other startups, including Ilya Sutskever 's $32 billion Safe Superintelligence , launched just a year ago. Sources told news channel CNBC that Meta tried to acquire the startup founded by Sutskever, former chief scientist at OpenAI, earlier this year. However, Sutskever rebuffed the big tech giant's efforts. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Play War Thunder now for free War Thunder Play Now Undo Subsequently, Zuckerberg started eyeing the chief executive of the startup, Daniel Gross. Zuckerberg is setting up a team of experts to achieve so-called "artificial general intelligence" (AGI), or machines that can match or surpass human capabilities. Live Events The Meta chief's apparent plans to personally recruit around 50 people, including a new head of AI research for the AGI team, is driven partly by frustration over the performance and reception of Meta's latest large language model, Llama 4. Discover the stories of your interest Blockchain 5 Stories Cyber-safety 7 Stories Fintech 9 Stories E-comm 9 Stories ML 8 Stories Edtech 6 Stories Altman swiped at Zuckerberg in his brother's podcast, 'Uncapped', saying that Meta tried to poach the company's AI talent for its superintelligence team. Additionally, he said Meta is not "great at innovation" . The 'special thing' about OpenAI is that the company has built a culture that is good at 'repeatable' innovation, Altman noted.
CNBC
2 days ago
- Business
- CNBC
The secret AI sauce behind Meta stock's 683% rise since the dark days of 2022
Picture scrolling through Facebook or Instagram and spotting an advertisement that feels tailor-made for you. You're seeing Meta Platform 's artificial intelligence tools in action, crafting ads for its partner businesses that specifically target and attract customers based on their interests. It has also become the secret sauce behind Meta's wildly success ad unit driving its stellar financial performance and stock performance. But few could see Meta's path to AI dominance just a few years ago. In 2022, the social media giant hit a low as investors balked at CEO Mark Zuckerberg's costly metaverse project and Apple's privacy changes disrupted its ad business. Meta shares fell by more than 60% in 2022, at one point closing as low as $88.91 a share. Then came Zuckerberg's "Year of Efficiency" in 2023 aimed at reversing the tide with layoffs and a focus on profitability. Meta quickly rewarded its loyal investors with a nearly 200% stock jump that year, as its AI-enhanced ads revived revenue growth and the company's cost cuts jumpstarted earnings. And despite lingering concerns over the company investing too much and too quickly in the nascent technology, it was hard to argue with the results. Meta delivered revenue and earnings beats for all four quarters of fiscal 2024. Jump to 2025 and Meta has fully repositioned itself as an AI-first company, one of the most celebrated in the field. The company's open-source Llama large language models are the cornerstone of its strategy, competing against the likes of OpenAI's GPT-4 and Google's Gemini. Zuckerberg's investments in data centers and hardware aim to secure the company's long-term edge. Building these large powerful AI models is critical to having "control of our own destiny" in powering the various AI opportunities Meta is focused on, Zuckerberg has said, emphasizing improved advertising and user experiences. As of Wednesday's close, its stock is up roughly 683% since that November 2022 closing low. Meta's AI strategy has not been without hurdles. The company delayed the release of Behemoth, its most advanced large language model, originally slated for April, until the fall or later due to performance issues. At the same time, Llama 4's release didn't generate much enthusiasm, fueling perceptions that Meta was falling behind in the AI race. The challenges compounded when 11 out of 14 AI researchers left the company amid intensified competition. The biggest risk to Meta's stock is losing its lead on the cutting edge of AI, said Gil Luria, analyst at D.A. Davidson, who noted that the underwhelming Llama 4 opened the door to competitors, including OpenAI, Anthropic and some Chinese models. Indeed, Zuckerberg personally stepped in to recruit top talent to stay out front, including bringing back Robert Fergus, a former Google DeepMind researcher and previous Meta employee, to enhance the AI division. Earlier this month, Meta made another major move investing $14.8 billion for a 49% stake in data-labeling company Scale AI and hired its CEO to lead a new "superintelligence" research lab, joined by Scale AI staff. In a sign of its aggressiveness, Meta also offered employees at OpenAI bonuses of $100 million to leave the ChatGPT creator, OpenAI CEO Sam Altman said on a recent podcast . Key growth ingredient So what exactly has made Meta's AI so special? The company's newest Llama 4 is "multimodal," which means it can process and learn from multiple types of data including text, images or backgrounds, similar to Google's Gemini and OpenAI's GPT-4 multimodal models. It powers Meta's ad growth by enabling advertisers to create tailored ads quickly and cost-effectively, and boosts engagement and performance across Facebook and Instagram. "Performance improvements from being multimodal benefit all applications," said Matt Steiner, Meta's VP of monetization infrastructure, ranking & AI foundations, in a CNBC interview. "Models trained on different data sources like images or videos or text all benefit from being trained on the same model," he added, noting this versatility drives better ad targeting and content creation. Steiner emphasized that this approach "helps us maintain our competitive edge in advertising by maximizing return on advertiser spend while controlling costs." Meta trails only Google in digital advertising, capturing 23% of global ad revenue in 2024 compared to Google's 28%, according to eMarketer. Meta's ad revenue rose 22% last year, outpacing Google's 12% and the industry's 9%. This growth, driven by AI-powered targeting, underscores Meta's momentum. "The ability to deliver better ads allows them to sell those ads for more, which is why they're gaining share in the digital ad market," said Luria, noting that Meta has been adept at using traditional machine learning techniques. He said Meta is taking this technology to an advanced level, "automating it further — allowing advertisers to generate content and make even more well-informed decisions." This next wave, generative AI, "deepens Meta's competitive moat." He highlightdc an example of how Reels is becoming more compelling to users. That's because Meta is "getting better at the AI algorithm that allows them to serve the next best short video to keep the consumer engaged." The introduction of ads on WhatsApp, announced this week, also expands the reach of Meta's AI ad tools, creating a new place where it can generate high-margin revenue. Meta on Tuesday announced new generative AI tools for its Advantage+ platform, enabling advertisers to integrate brand elements into personalized ads and create animated videos from images with music and text overlays. The company is also testing a new feature in "Video Highlights" which uses AI summaries to better digest video ads by skipping to the highlights of the video. Looking ahead, Meta's generative AI advancements are enabling it to deliver a higher volume of personalized ads. The goal: To keep and strengthen its AI edge. As Meta's Steiner put it, the company's "compounding effects" of AI improvements fuel growth, which should help Meta make more gains in digital advertising. The company's financial and strategic commitment show Zuckerberg and Co. are in a league of their own in digital ads as Meta continues to take market share. We're confident in Meta's ability to keep innovating on its advertising tools using AI, keeping customers around and its business growing. We currently have a 2 rating on the stock with a price target of $750. (Jim Cramer's Charitable Trust is long META. See here for a full list of the stocks.) As a subscriber to the CNBC Investing Club with Jim Cramer, you will receive a trade alert before Jim makes a trade. Jim waits 45 minutes after sending a trade alert before buying or selling a stock in his charitable trust's portfolio. If Jim has talked about a stock on CNBC TV, he waits 72 hours after issuing the trade alert before executing the trade. THE ABOVE INVESTING CLUB INFORMATION IS SUBJECT TO OUR TERMS AND CONDITIONS AND PRIVACY POLICY , TOGETHER WITH OUR DISCLAIMER . NO FIDUCIARY OBLIGATION OR DUTY EXISTS, OR IS CREATED, BY VIRTUE OF YOUR RECEIPT OF ANY INFORMATION PROVIDED IN CONNECTION WITH THE INVESTING CLUB. NO SPECIFIC OUTCOME OR PROFIT IS GUARANTEED.
Time of India
3 days ago
- Business
- Time of India
OpenAI CEO Sam Altman reveals why ChatGPT maker's 'best people' rejected Mark Zuckerberg's $100 million offers: They're a company that's ...
OpenAI CEO Sam Altman delivered a brutal assessment of Meta 's innovation capabilities, explaining why his top engineers turned down Mark Zuckerberg 's eye-watering $100 million signing bonuses. "There's many things I respect about Meta as a company, but I don't think they're a company that's great at innovation," Altman said on his brother's "Uncapped" podcast Tuesday, revealing the core reason behind his staff's rejection of the offers. Speaking candidly about Meta's aggressive poaching attempts, Altman said the Facebook founder had "started making these giant offers to a lot of people on our team" but boasted that "none of our best people have decided to take them up on that." Why OpenAI engineers chose Sam Altman's OpenAI over working for Mark Zuckerberg The OpenAI chief suggested his employees see the company as having "a much better shot actually, delivering on superintelligence and also may eventually be the more valuable company." Despite Meta's $1.77 trillion market cap dwarfing OpenAI's $300 billion valuation, the ChatGPT maker's engineers appear unconvinced by pure financial incentives. Altman criticized Meta's strategy of focusing on massive upfront compensation packages, saying it would damage company culture. "The degree to which they are focusing on that and not the work and not the mission, I don't think that's going to set up a great culture," he said. Meta's desperate talent hunt amid AI setbacks Meta's aggressive recruitment drive comes after a series of embarrassing setbacks in the AI race. The company has delayed its flagship "Behemoth" AI model and faced criticism over inflated performance metrics for its Llama 4 language model. "I've heard that Meta thinks of us as their biggest competitor," Altman said. "Their current AI efforts have not worked as well as they have hoped and I respect being aggressive and continuing to try new things." The Facebook founder has been personally calling AI researchers as part of his frantic bid to build a "superintelligence" team, even rearranging Meta's Menlo Park headquarters to put the new unit near his office. Meta recently invested $14.3 billion in Scale AI and hired its founder Alexandr Wang , but the talent war for AI supremacy rages on. AI Masterclass for Students. Upskill Young Ones Today!– Join Now
International Business Times
3 days ago
- International Business Times
Today's AI Updates: OpenAI Brings Image Creation to WhatsApp As Microsoft Rolls Out Smart AI Agent for Windows PCs
As the race to lead the artificial intelligence boom intensifies, top technology firms are leaving no stone unturned in their efforts to outpace competitors by offering innovative features and services. In the latest wave of AI developments, OpenAI, Meta, and Microsoft have introduced distinctive new features for their users. Starting with OpenAI, the Sam Altman-led firm has officially launched its image-generation tool on WhatsApp today, allowing users to create pictures within the messaging app using text-based instructions Using OpenAI's DALL·E model, the feature allows users to message 'Hi' to the official number +1-800-242-8478, register using their official mobile number, and generate AI images in real-time by following simple steps and prompts. This eliminates the need to manually switch to ChatGPT and creates a more seamless experience within WhatsApp. Another tech leader, Meta, is also offering a similar feature. Meta's own AI assistant, powered by Llama 4, is providing image generation and conversational skills within WhatsApp. Meta's AI can be used to suggest writing prompts, edit images and videos, and compose responses to user queries. Meanwhile, Microsoft is adding a new AI settings agent to Windows 11 as part of its Copilot platform, starting with Copilot+ PCs with Snapdragon X processors. The tool is a departure from traditional methods for interacting with desktop settings (obtained through laborious multi-level menus), in that users communicate with it using succinct, natural commands. The AI agent allows users to complete actions like adjusting brightness, altering contrast, or connecting to Wi-Fi by typing or speaking conversational commands, such as "turn on dark mode" or "make text easier to read." In contrast with old digital assistants, the new AI agent has the ability to not just interpret user commands but also execute real-time actions (with user confirmation), regardless of how they are expressed. Microsoft stated that these interactions are performed locally on the device for better privacy and also faster response of the bot. The agent's screen awareness enables it to know what the user is working on and offer contextual help. If someone is editing a photo, for example, it could suggest improvements. When viewing a PDF, it can summarize the content of that document. Microsoft says this approach is aimed at doing away with artificial frictions and making Windows feel like a more responsive, intelligent environment. While only available on Copilot+ PCs for now, Microsoft intends to bring it to all Intel and AMD systems later this year when it releases the Windows 11 24H2 update. The company also mentioned early on that it would eventually allow third-party developers to build their own AI agents using its Model Context Protocol, which could make it possible to coordinate across apps and integrate more artificial intelligence throughout Windows.
