Latest news with #GoogleDeepMind
India Today
2 days ago
- Business
- India Today
Immigrants powering US AI push: From Bansal to Chang, Zuckerberg's top AI team is full of immigrants
As the AI arms race between Silicon Valley giants intensifies, Meta CEO Mark Zuckerberg is upping the ante with a bold new move — the creation of a brand-new SuperIntelligence Lab and a star-studded lineup of hires to go with it. In what can only be described as a talent heist, Meta has hired 11 of the brightest minds in AI, poaching top talent from OpenAI, Google DeepMind, Anthropic and beyond. Each new recruit brings high-end expertise to the table, from multimodal systems to LLM safety. But it is interesting to see how none of them hold a bachelor's degree from the United other words, Meta's SuperIntelligence Lab team — and in general almost entirety of the US AI push — is powered by immigrants, the same people who are currently on the radar of the Donald Trump administration. So, who are these people? Let's meet Meta's newest brain trust:Trapit BansalBefore joining the Meta team, Trapit Bansal was one of the OpenAI brains. An alumnus of IIT Kanpur, he holds integrated degrees in mathematics and statistics, followed by a PhD in computer science from the University of Massachusetts where he focused on meta-learning, deep learning, and natural language processing. His professional journey blends strong academic foundations with practical experience at some of the world's leading tech Bi Shuchao Bi, formerly of OpenAI, brings his multimodal magic to Meta's SuperIntelligence Lab. A top expert in blending speech and text, Bi played a key role in developing GPT-4o's voice mode and the compact yet powerful o4-mini model. His work bridges the gap between how humans communicate and how AI understands, making digital assistants more fluid, natural, and versatile than ever. From boosting conversational AI to shaping the future of human-machine interaction, Bi's expertise is set to be a game-changer for Meta's AI ChangPreviously at Google Research, Chang, is a known generative image wizard. She helped design the Muse and MaskIT architectures, the secret sauce behind many AI-generated visuals today. She studied at Tsinghua University in China, and then in Princeton. She also led the image generation work for LinLin is another OpenAI talent who is a Tsinghua and MIT alumnus. He was instrumental in scaling LLMs like GPT-4o. His work has made high-quality AI image generation more efficient and cost-effective — a major win for large-scale PobarAfter earning his degree in Australia, Joel Pobar carved out a name for himself as a seasoned infrastructure expert with more than a decade of experience in building scalable AI systems. His work on industry-defining projects like HHVM, Hack, and PyTorch has made him a key player behind the scenes of modern AI. Jack RaeA former employee at Google DeepMind, Rae has a reputation as a language model heavyweight. He's been involved in pre-training major models like Gemini 2.5, Gopher and Chinchilla. With degrees from Bristol, CMU, and UCL, he brings a unique cross-Atlantic academic RenRen, a Stanford PhD and Peking University grad, worked on post-training GPT-4o and its smaller variants. His research is focused on improving AI reliability, making models safer, more robust, and generally less prone to hallucinations. He is the fourth ex-OpenAI SchalkwykadvertisementA former Googler and a South African speech recognition expert, Schalkwyk helped lead the Maya team and contributed to the early days of the Sesame project. He's now bringing his voice tech expertise to Meta's AI SunSun's dual degrees from Tsinghua and CMU power his work in post-training advanced AI models. He's also built perception systems for Waymo's autonomous vehicles, giving him a rare edge in real-world AI deployment. He quit Google DeepMind to come onboard with Meta. Jiahuai YuA multimodal maverick, Yu has contributed to a range of powerful models like o3/4o-mini and GPT-4. With degrees from USTC and UIUC, his work lets AI systems understand and reason across text, images and ZhaoZhao is the co-creator of ChatGPT, GPT-4 and o4-mini. A leader in data synthesis and AI safety, Zhao's research continues to shape how modern AI models learn and behave responsibly.- Ends
Gulf Insider
3 days ago
- Entertainment
- Gulf Insider
Veo 3: Google's Next-Gen AI Video Tool Launches On Gemini Across Middle East
Now paying Gemini users across the Middle East can use Veo 3, Google's next-generation AI-powered video creation tool that transforms simple text prompts into rich, cinematic video clips complete with sound, music, dialogue and more. Announced during Google I/O, the tech giant's annual developer event in May, Veo 3 stands out for its native audio generation, covering everything from voiceovers to background noise, which puts it in direct competition with rivals like OpenAI's Sora. Users can describe any scene they imagine in a short prompt, and the AI will generate a 720p, eight-second video featuring synchronised audio and visuals that include ambient sound, character dialogue, and even realistic sound effects. Unveiled at Google I/O in May, 'From prompt to production, Veo 3 delivers best-in-class realism, physics, and lip syncing,' said Eli Collins, VP of Product at Google DeepMind. Whether it's a dreamlike short film or a viral meme, Veo 3 is already making waves online. Unlike other models, 'Veo 3 excels from text and image prompting to real-world physics and accurate lip syncing,' said Eli Collins, VP of Product at Google DeepMind, in a blog post highlighting the model's strengths in realism, responsiveness, and user control. All videos generated by Veo 3 come with a hidden SynthID watermark, Google's digital signature to label AI-generated content. The company is also testing a SynthID Detector tool to help identify synthetic media. Additionally, all Veo-generated clips – except those made by Ultra-tier members using Google's new Flow filmmaking platform – will carry a visible watermark for transparency. Since its release elsewhere in the world, creators have flocked to platforms like X (formerly Twitter), where Veo 3 has been used to generate everything from animated short films to viral meme content – including a surreal clip of Will Smith eating spaghetti.
Time of India
4 days ago
- Business
- Time of India
Google, you broke your word on …, shout protestors outside Google Deepmind's London headquarters
Dozens of protesters staged a mock courtroom trial outside Google DeepMind 's London headquarters Monday, accusing the AI giant of breaking public safety promises made during the launch of its Gemini 2.5 Pro model. Tired of too many ads? go ad free now The demonstration, organized by activist group PauseAI , drew over 60 participants who chanted "Test, don't guess" and "Stop the race, it's unsafe" while conducting a theatrical trial complete with a judge and jury. The group claims violated commitments made at the 2024 AI Safety Summit in Seoul, where the company pledged to involve external evaluators in testing its advanced AI models and publish detailed transparency reports. When Google released Gemini 2.5 Pro in April, it labeled the model "experimental" and initially provided no third-party evaluation details. A safety report published weeks later was criticized by experts as lacking substance and failing to identify external reviewers. Companies less regulated than sandwich shops, says protestors "Right now, AI companies are less regulated than sandwich shops," said PauseAI organizing director Ella Hughes, addressing the crowd. "If we let Google get away with breaking their word, it sends a signal to all other labs that safety promises aren't important." The protest reflects growing public concern about AI development pace and oversight. PauseAI founder Joep Meindertsma , who runs a software company and uses AI tools from major providers, said the group chose to focus on this specific transparency issue as an achievable near-term goal. Monday marked PauseAI's first demonstration targeting this particular Google commitment. The group is now engaging with UK Parliament members to escalate their concerns through political channels. Google has not responded to requests for comment about the protesters' demands or future transparency plans for its AI models.
The Hindu
4 days ago
- Business
- The Hindu
Meta poaches top AI talent from OpenAI and DeepMind as Zuckerberg escalates AI push
The minds that brought you the conversational magic of ChatGPT and the multimodal power of Google's Gemini now have a new home: Meta. In a stunning talent exodus, captured in a single tweet by Meta's new AI chief, Alexandr Wang, the architects of the current AI revolution have been poached from OpenAI, Google DeepMind, and Anthropic. Mr. Alexandr's tweet is not merely a hiring announcement; it is a declaration of intent. By announcing his role as Chief AI Officer at Meta alongside Nat Friedman and a veritable 'who's who' of top-tier AI researchers, Mr. Alexandr and Meta were signaling a seismic shift in the technology landscape. This mass talent acquisition from rivals like OpenAI, Google DeepMind, and Anthropic is CEO Mark Zuckerberg's most audacious move yet to dominate the next technological frontier. It represents a calculated effort to bolster Meta's AI venture by poaching the very minds that built its competitors' greatest successes. However, this aggressive pivot towards superintelligence cannot be viewed in a vacuum. It is haunted by the ghosts of Meta's past — from the Cambridge Analytica scandal to the Instagram teen mental health crisis — forcing a critical examination of whether Zuckerberg has evolved from a disruptive force into a responsible steward for the age of AI. Why Meta's AI talent acquisition signals a new era The list of new hires is a strategic masterstroke — it's not just about adding headcount; it's about acquiring institutional knowledge and simultaneously weakening the competition. According to reports, Mr. Zuckerberg has personally handled these AI hires. And he has carefully picked top talent from all his rivals. From OpenAI, Meta has poached the creators behind GPT-4o's groundbreaking voice and multimodal capabilities and foundational model builder. Shengjia Zhao, the co-creator of ChatGPT and GPT-4, is now part of Meta. This is a significant loss to Sam Altman's AI company. From Google DeepMind, Mr. Zuckerberg has poached Jack Rae, the pre-training tech lead for Gemini 2.5, and other experts in the text-to-image generation. From Anthropic, Meta has poached Joel Pobar, the AI firm's inference expert. This talent raid provides Meta with some immediate advantages. First, it gives the company instant credibility that it quite serious about its AI bet as the new team has direct, hands-on experience building and training the world's most advanced models. Second, it disrupts the roadmaps of its competitors, forcing them to regroup and replace key personnel. Third, it creates a powerful gravitational pull for future talent, signaling that Meta is now the premier destination for ambitious AI work, backed by near-limitless computational resources and a direct path to impacting billions of users. Can Zuckerberg be trusted with the future of AI? This aggressive push into AI stands in stark contrast to the defining scandals of Zuckerberg's career. The Cambridge Analytica affair revealed a fundamental flaw in Facebook's DNA: a platform architecture that prioritized growth and data collection over user privacy and security, which was then exploited for political manipulation. The company's response was slow, defensive, and ultimately insufficient to repair the deep chasm of public trust. Then, 'The Facebook Files' exposé by The Wall Street Journal detailed internal research showing that Meta knew Instagram was toxic for the mental health of teenage girls. The company's leadership chose to downplay the findings and continue with product strategies that exacerbated these harms. Both incidents stem from the same root philosophy: 'move fast and break things,' a mantra that prioritizes scale and engagement above all else, with societal consequences treated as unfortunate but acceptable collateral damage. Applying this ethos to AI, a technology with far greater potential for both good and harm, is a terrifying prospect. If a social feed algorithm could destabilize democracies and harm teen self-esteem, what could a superintelligent agent, deployed to three billion users with the same growth-at-all-costs mindset, be capable of? Mr. Zuckerberg's past misadventures are not just historical footnotes; they are the core reason for public skepticism towards Meta's AI ambitions. How Zuckerberg has evolved from social media to superintelligence Mr. Zuckerberg's character, as observed through his actions over two decades, is one of relentless, almost singular, ambition. He has consistently demonstrated a willingness to be ruthless in competition (cloning Snapchat's features into Instagram Stories), a visionary in long-term bets (acquiring Instagram and WhatsApp, pivoting to the Metaverse), and an ability to withstand immense public and regulatory pressure. His critics would argue he is a leader who lacks a deep-seated ethical framework, often optimizing for power and market dominance while retroactively applying ethical patches only when forced by public outcry. His defenders might say he is a pragmatic engineer who is learning and adapting. The Cambridge Analytica scandal arguably forced him to mature from a hoodie-wearing coder into a global CEO who must at least speak the language of governance and responsibility. How Meta's AI super-team challenges OpenAI and Google The crucial question is whether this change is superficial or substantive. His current strategy with AI suggests a potential evolution. The open-sourcing of the Llama models can be interpreted in two ways. On one hand, it's a shrewd business move to commoditise the layer of the stack where OpenAI and Google have a strong lead, fostering an ecosystem dependent on Meta's architecture. On the other, it can be framed as a commitment to transparency and democratisation, a direct response to the 'black box' criticism leveled at his past operations. This new 'super-team' will be the ultimate test. Will they be fire-walled by a new ethical charter, or will the immense pressure from Mr. Zuckerberg to 'win' the AI race override all other considerations? How is Meta positioning itself for the AI age Against the closed, API-first models of OpenAI and the integrated-but-cautious approach of Google, Meta is carving out a unique strategic position. It is fighting the war on two fronts — by making Llama an open-source alternative, Meta is making itself the default foundation for thousands of startups, researchers, and developers, disrupting the business models of its rivals. Mr. Zuckerberg hasn't stopped with that, he has also publicly committed to acquiring hundreds of thousands of high-end NVIDIA GPUs, signaling that his company will not be outspent on compute. With the addition of this new team, Meta completes the trifecta: massive data, unparalleled compute, and now, world-leading human talent. The goal is no longer just to build a chatbot for Messenger or an image generator for Instagram. As Mr. Alexandr's tweet boldly states, the aim is 'Towards superintelligence.' This is a direct challenge to the stated missions of DeepMind and OpenAI. The formation of this AI super-team is the culmination of Mr. Zuckerberg's pivot from social media king to aspiring AI emperor. It is an act of immense strategic importance, one that immediately elevates Meta to the top tier of AI development. Yet, the success of this venture will not be measured solely by the capability of the models it produces. It will be measured by whether Mr. Zuckerberg can build an organization that has learned from the profound societal failures of its past. This is a defining gambit for Meta founder — a chance to redefine his legacy not as the creator of a divisive social network, but as the leader who responsibly ushered in the age of artificial intelligence.
Time of India
4 days ago
- Health
- Time of India
Kilkari and mMitra: How Google DeepMind is helping improve maternal health
ARMMAN , a Mumbai-based non-profit focused on maternal health , is leveraging AI with pro bono support from Google DeepMind to enhance its outreach programs, Kilkari and mMitra , and is said to be yielding promising results in improving maternal and child health across India. Kilkari, the world's largest mobile-based maternal health program, partners with India's Ministry of Health to deliver weekly audio messages on pregnancy, childbirth, and childcare to over 60 million women across 27 states. These free calls, starting in the second trimester and continuing until a year post-delivery, aim to reduce healthcare lapses. However, a 23% drop-off rate prompted ARMMAN to collaborate with Google DeepMind to develop an AI model that predicts optimal call times by analyzing anonymized individual and group call behaviors. A pilot in Odisha with 17,500 participants showed a 12% increase in call pickup rates for certain time slots, enhancing the delivery of critical health information. Similarly, ARMMAN's mMitra program, serving 350,000 women in Maharashtra, used AI to identify participants at risk of disengaging. By prioritizing them for personalized outreach, including in-person support, the program retained 30% of high-risk participants. A health impact assessment revealed significant improvements: women prioritized by the AI were 22% more likely to take iron supplements, 28% more likely to take calcium, and 9% better at tracking their babies' birth weights. These programs also foster social change, empowering women to advocate for better healthcare and rights for their daughters. ARMMAN's AI-driven approach, replacing assumptions with data-driven insights, demonstrates technology's potential to address global health challenges. With plans for nationwide expansion and open-sourced models, ARMMAN aims to inspire similar innovations worldwide, proving AI's role as an accelerant for progress in maternal health.
