Latest news with #NeurIPS
Time of India
03-08-2025
- Business
- Time of India
Who is Matt Deitke? The 24-year-old AI researcher and PhD dropout behind Meta's $250 million offer
In an era where artificial intelligence is rapidly redrawing the boundaries of human capability, the competition to secure the brightest minds has intensified into an all-out global race. Amidst billion-dollar investments, multimodal breakthroughs, and the relentless pursuit of artificial general intelligence, one name has recently emerged as the embodiment of this new wave of AI ambition, Matt Deitke. At just 24, Deitke has done what few in any field can claim: Walked away from a prestigious PhD, co-founded a startup at the edge of AI autonomy, and turned down a nine-figure job offer, only to see it doubled by one of the most powerful tech CEOs on the planet. This is not a story of overnight success. It's the tale of a young researcher whose talent, timing, and tenacity are rewriting the rules of how AI careers unfold, and whose trajectory now sits at the intersection of cutting-edge science, billion-dollar bets, and the strategic future of Big Tech. The scholar who walked away Matt Deitke's early career followed a familiar path for a rising academic star. As a doctoral student in computer science at the University of Washington, he immersed himself in a field undergoing seismic change. But where others saw a traditional climb through academia, Deitke sensed the urgency of a moment that could not be paused. Rather than remain in the ivory tower, he chose to engage directly with the frontier. He left the PhD program, an unconventional but increasingly common decision among elite AI researchers, and joined the Allen Institute for Artificial Intelligence (AI2) in Seattle, a renowned research hub founded by Microsoft co-founder Paul Allen. There, he didn't just contribute; he led. Deitke spearheaded the development of Molmo, a chatbot built not only to process text but to understand images and audio, ushering in a more human-like form of machine understanding. This multimodal capacity represents one of the most important advances in AI today, and Deitke was already at its core. Recognition and reinvention Deitke's work quickly caught the attention of the global AI community. At NeurIPS 2022, one of the most prestigious conferences in machine learning, he received an Outstanding Paper Award, an accolade that signals a researcher's arrival on the world stage. But Deitke wasn't content with accolades. In 2023, he co-founded Vercept, a startup focused on building autonomous AI agents that don't just interpret the web, but navigate it and act within it. The idea was radical: Systems that can take goals and execute tasks across the internet, mimicking the autonomy of human behaviour in digital environments. The startup, though lean with just ten team members, raised $16.5 million from a high-profile group of investors that included Eric Schmidt, the former CEO of Google. Vercept represents the vanguard of where AI is headed, beyond chatbots and recommendation engines, toward agents capable of real-world digital action. And at its helm was a 24-year-old who had already turned down one of the biggest job offers in tech history. Meta 's $250 million bet When Meta first approached Deitke with an offer reportedly worth $125 million over four years, it was already a headline-making move. But in a dramatic twist, Deitke declined. That rejection prompted a personal meeting with Mark Zuckerberg, who made a counter offer that stunned even seasoned Silicon Valley observers: $250 million. The deal, among the most generous compensation packages ever extended to a researcher of any age, was emblematic of Meta's increasingly aggressive AI recruitment strategy. It recently onboarded Ruoming Pang, the former leader of Apple's AI models team, in a package reportedly exceeding $200 million. In 2025 alone, Meta is expected to spend $72 billion on capital expenditure—including massive investments in compute infrastructure and AI talent. A new model for AI careers Matt Deitke's story is more than a tale of youth and fortune; it's a parable for the new reality of AI. The boundaries between academia, industry, and entrepreneurship are no longer rigid. In fact, they are dissolving. Researchers now operate in a landscape where intellectual achievement can translate into unprecedented wealth, influence, and impact. Yet Deitke's choices reflect more than opportunism. They show strategic clarity. Rather than lock himself into a single institution or trajectory, he has navigated the ecosystem with autonomy, mirroring the very kind of AI he seeks to build. At 24, Matt Deitke stands not only as a prodigy but as a prototype: The kind of polymath-entrepreneur-researcher hybrid that today's AI revolution demands. Whether at Vercept or Meta, his work will likely shape the tools, agents, and intelligence systems that define the coming decade. As Silicon Valley, academia, and the global tech community look to the future of artificial intelligence, one thing is increasingly clear: Matt Deitke isn't just along for the ride; he's driving the evolution. Ready to navigate global policies? Secure your overseas future. Get expert guidance now!
The Guardian
03-08-2025
- Business
- The Guardian
What will the AI revolution mean for the global south?
I come from Trinidad and Tobago. As a country that was once colonized by the British, I am wary of the ways that inequalities between the global north and global south risk being perpetuated in the digital age. When we consider the lack of inclusion of the global south in discussions about artificial intelligence (AI), I think about how this translates to an eventual lack of economic leverage and geopolitical engagement in this technology that has captivated academics within the industrialised country I reside, the United States. As a scientist, I experienced an early rite of passage into the world of Silicon Valley, the land of techno-utopianism, and the promise of AI as a net positive for all. But, as an academic attending my first academic AI conference in 2019, I began to notice inconsistencies in the audience to whom the promise of AI was directed. AI researchers can often identify consistent choices for locations where such conferences are hosted, and where they are not. NeurIPS, one of the top AI conferences, has highlighted annual issues for obtaining visas for academic attendees and citizens from the African continent. Attending such a prestigious conference in the field grants one the opportunity to gain access to peers in the field, new collaborations and feedback on one's work. I often hear the word 'democratisation' within the AI community, an implication of equity in access, opportunity and merit for contribution regardless of one's country of origin. Associate professor of economics Fadhel Kaboub talks about how 'a lack of vision for oneself results in being a part of someone else's vision', reflecting on how systematically lacking an access to infrastructure results in local trade deficits in economies. As in the time of Nafta's promise of 'free trade', promises of 'AI democratisation' today still exist and benefit mainly countries with access to tech hubs not located in the global south. While the United States and other industrialized countries dominate in access to computational power and research activity, much of the low-paid manual labour involved in labelling data and the global underclass in artificial intelligence still exists in the global south. Much like coffee, cocoa, bauxite and sugar cane are produced in the global south, exported cheaply and sold at a premium in more industrialized countries, over the past few years we have seen influence in AI inextricably tied to energy consumption. Countries that can afford to consume more energy have more leverage in reinforcing power to shape the future direction of AI and what is considered valuable within the AI academic community. In 2019, Mary L Gray and Siddharth Suri published Ghost Work, which exposed the invisible labour of technology today, and at the beginning of my tenure at graduate school, the heavily cited paper Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence was published. It has been five years since these seminal works. What would an AI community inspired by the Brics organisation, which united major emerging economies to advocate for themselves in a system dominated by western countries, look like for the global south? I often ask myself how AI has contributed to our legacy, and whose stories it won't tell. Has AI mitigated issues of mistrust and corruption in less-resourced countries? Has it benefited our civic communities or narrowed educational gaps between less-resourced regions? How will it make society better, and whose society will it make better? Who will be included in that future? A historical mistrust can impede adoption by developing countries. Furthermore, many developing countries have weak institutional infrastructures, poor or nonexistent laws and regulatory frameworks for data projection and cybersecurity. Therefore, even with an improved information infrastructure, they are likely to function at a disadvantage in the global information marketplace. A currency is only as good as its perceived global trust. When thinking about the democratization in AI and a vision of what it could be in years to come, AI's survival requires including more perspectives from regions such as the global south. Countries from the global south should work together to build their own markets and have a model of sovereignty for their data and data labour. Economic models often consider a definition of development that includes a measure of improvement in the quality of life of the most marginalized of its people. It is my hope that in the future that will extend to our evaluation of AI. Krystal Maughan is a PhD student at the University of Vermont studying differential privacy and machine learning
The Guardian
03-08-2025
- Business
- The Guardian
What will the AI revolution mean for the global south?
I come from Trinidad and Tobago. As a country that was once colonized by the British, I am wary of the ways that inequalities between the global north and global south risk being perpetuated in the digital age. When we consider the lack of inclusion of the global south in discussions about artificial intelligence (AI), I think about how this translates to an eventual lack of economic leverage and geopolitical engagement in this technology that has captivated academics within the industrialised country I reside, the United States. As a scientist, I experienced an early rite of passage into the world of Silicon Valley, the land of techno-utopianism, and the promise of AI as a net positive for all. But, as an academic attending my first academic AI conference in 2019, I began to notice inconsistencies in the audience to whom the promise of AI was directed. AI researchers can often identify consistent choices for locations where such conferences are hosted, and where they are not. NeurIPS, one of the top AI conferences, has highlighted annual issues for obtaining visas for academic attendees and citizens from the African continent. Attending such a prestigious conference in the field grants one the opportunity to gain access to peers in the field, new collaborations and feedback on one's work. I often hear the word 'democratisation' within the AI community, an implication of equity in access, opportunity and merit for contribution regardless of one's country of origin. Associate professor of economics Fadhel Kaboub talks about how 'a lack of vision for oneself results in being a part of someone else's vision', reflecting on how systematically lacking an access to infrastructure results in local trade deficits in economies. As in the time of Nafta's promise of 'free trade', promises of 'AI democratisation' today still exist and benefit mainly countries with access to tech hubs not located in the global south. While the United States and other industrialized countries dominate in access to computational power and research activity, much of the low-paid manual labour involved in labelling data and the global underclass in artificial intelligence still exists in the global south. Much like coffee, cocoa, bauxite and sugar cane are produced in the global south, exported cheaply and sold at a premium in more industrialized countries, over the past few years we have seen influence in AI inextricably tied to energy consumption. Countries that can afford to consume more energy have more leverage in reinforcing power to shape the future direction of AI and what is considered valuable within the AI academic community. In 2019, Mary L Gray and Siddharth Suri published Ghost Work, which exposed the invisible labour of technology today, and at the beginning of my tenure at graduate school, the heavily cited paper Decolonial AI: Decolonial Theory as Sociotechnical Foresight in Artificial Intelligence was published. It has been five years since these seminal works. What would an AI community inspired by the Brics organisation, which united major emerging economies to advocate for themselves in a system dominated by western countries, look like for the global south? I often ask myself how AI has contributed to our legacy, and whose stories it won't tell. Has AI mitigated issues of mistrust and corruption in less-resourced countries? Has it benefited our civic communities or narrowed educational gaps between less-resourced regions? How will it make society better, and whose society will it make better? Who will be included in that future? A historical mistrust can impede adoption by developing countries. Furthermore, many developing countries have weak institutional infrastructures, poor or nonexistent laws and regulatory frameworks for data projection and cybersecurity. Therefore, even with an improved information infrastructure, they are likely to function at a disadvantage in the global information marketplace. A currency is only as good as its perceived global trust. When thinking about the democratization in AI and a vision of what it could be in years to come, AI's survival requires including more perspectives from regions such as the global south. Countries from the global south should work together to build their own markets and have a model of sovereignty for their data and data labour. Economic models often consider a definition of development that includes a measure of improvement in the quality of life of the most marginalized of its people. It is my hope that in the future that will extend to our evaluation of AI. Krystal Maughan is a PhD student at the University of Vermont studying differential privacy and machine learning
Time of India
02-07-2025
- Science
- Time of India
How to Become an AI Genius: Lessons students can learn from Meta's $100 million hires
If you want to become an AI genius – the kind that Mark Zuckerberg offers $50–$100 million to join his quest for artificial general intelligence (AGI) – here's the blueprint, decoded from Meta's elite hires. 1. Build a rock-solid maths foundation Almost every AI superstar Meta poached – from Lucas Beyer to Trapit Bansal – started with hardcore mathematics or computer science degrees. Linear algebra, calculus, probability, and optimisation aren't optional. They are your bread and butter. Why? Because AI models are just giant stacks of matrix multiplications optimised over billions of parameters. If you can't handle eigenvectors or gradient descent, you'll be stuck fine-tuning open-source models instead of inventing the next GPT-5. 2. Specialise in deep learning Next comes deep learning mastery. Study neural networks, convolutional networks for vision, transformers for language, and recurrent models for sequence data. The Vision Transformer (ViT) co-created by Lucas Beyer and Alexander Kolesnikov redefined computer vision precisely because they understood both transformer architectures and vision systems deeply. Recommended learning path: Undergraduate/early coursework : Machine learning, statistics, data structures, algorithms. Graduate-level depth : Neural network architectures, representation learning, reinforcement learning. 3. Research, research, research The real differentiator isn't coding ability alone. It's original research. Look at Meta's dream team: Jack Rae did a PhD in neural memory and reasoning. Xiaohua Zhai published groundbreaking papers on large-scale vision transformers. Trapit Bansal earned his PhD in meta-learning and reinforcement learning at UMass Amherst before co-creating OpenAI's o-series reasoning models. Top AI labs hire researchers who push knowledge forward, not just engineers who implement existing algorithms. This means: Reading papers daily (Arxiv sanity or Twitter AI circles help). Writing papers for conferences like NeurIPS, ICML, CVPR, ACL. 4. Dive into multimodal and reasoning systems If you want to be at the AGI frontier, focus on multimodal AI (vision + language + speech) and reasoning/planning systems. by Taboola by Taboola Sponsored Links Sponsored Links Promoted Links Promoted Links You May Like Glicemia acima de 100? Insira essa fruta na sua dieta Saúde Nacional Undo Why? Because AGI isn't just about language models completing your sentences. It's about: Understanding images, videos, and speech seamlessly Performing logical reasoning and planning over long contexts For example, Hongyu Ren's work combines knowledge graphs with LLMs to improve question answering. Jack Rae focuses on LLM memory and chain-of-thought reasoning. This is the cutting edge. 5. Optimise your engineering skills Finally, remember that AI breakthroughs don't live in papers alone. They need to run efficiently at scale. Pei Sun and Joel Pobar are prime examples: engineering leaders who ensure giant models run on hardware without melting the data centre. Learn: Distributed training frameworks (PyTorch, TensorFlow) Systems optimisation (CUDA, GPUs, AI accelerators) Software engineering best practices for scalable deployment The bottom line Becoming an AI genius isn't about quick YouTube tutorials. It's about mastering mathematics, deep learning architectures, original research, multimodal reasoning, and scalable engineering. Do this, and maybe one day, Mark Zuckerberg will knock on your door offering you a $50 million signing bonus to build his artificial god. Until then, back to those linear algebra problem sets. The future belongs to those who understand tensors. Is your child ready for the careers of tomorrow? Enroll now and take advantage of our early bird offer! Spaces are limited.
Yahoo
28-05-2025
- Business
- Yahoo
Quantiphi Unveils Phi Labs Research and Development Hub, Accelerating Breakthroughs in Generative AI, Life Sciences, Digital Twins
MARLBOROUGH, Mass., May 28, 2025 /PRNewswire/ -- Quantiphi, an AI-first digital engineering company, today unveiled Phi Labs by Quantiphi, the formal identity of Quantiphi's long-standing Research and Development (R&D) hub, addressing real-world business challenges through breakthrough AI solutions. Quantiphi's Research and Development has been at the forefront of innovation in Generative AI, Life Sciences and Digital Twins since 2018, with a patent portfolio that underscores its commitment to future-proofing its technology. Now known as Phi Labs by Quantiphi, the R&D hub's pioneering work has been featured in leading forums like Neural Information Processing Systems (NeurIPS) and the International Conference on Machine Learning (ICML). "Phi Labs embodies Quantiphi's vision to redefine AI's business impact," Quantiphi Global R&D Head Dagnachew (Dan) Birru said. "By integrating cutting-edge research with practical applications, we enable our clients to navigate the evolving AI landscape with confidence." Phi Labs' key achievements include: Technology and IP Transfers: More than 15 transfers leading to accelerators and product launches like baioniq, Dociphi and Codeaira Innovative Milestones: Projects ranging from Speech-to-Text for Finance, 3D-GANs for seismic imaging and advanced healthcare diagnostics, achieving up to a 90 percent reduction in processing time Recent Developments: From 2021 to 2024, Quantiphi filed 50-plus patents—including 30-plus by Phi Labs, secured multiple technology transfers and earned accolades through academic collaborations and publications, including being featured in the Financial Times. Phi Labs by Quantiphi focuses on: Generative AI: Advanced reasoning, agentic workflows, efficient models and intelligent document processing Life Sciences: Small molecule synthesis, genomics, protein engineering and pharmacovigilance Digital Twins: Operations optimization, digital twin agents, engineering simulation and climate risk modeling Quantum Computing: Quantum machine learning, hybrid quantum-classical systems "Unlike labs driven by theory, our approach embeds innovation into enterprise environments directly, while upholding Responsible AI standards," Birru said. "Our engineering-driven team pairs deep research with industry expertise, ensuring every solution delivers tangible business impact." About Quantiphi:Quantiphi is an award-winning AI-first digital engineering company driven by the desire to reimagine and realize transformational opportunities at the heart of business. Since its inception in 2013, Quantiphi has solved the toughest and most complex business problems by combining deep industry experience, disciplined cloud and data engineering practices, and cutting-edge artificial intelligence research to achieve accelerated and quantifiable business results. Learn more at and follow us on LinkedIn, X, formerly Twitter, and Instagram and YouTube. Media Contact:H. Newsroom View original content to download multimedia: SOURCE Quantiphi Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data
